Effects of malaria double infection in birds: one plus one is not two


A. Marzal, Department of Animal Ecology, Ecology Building, Lund University, S-223 62 Lund, Sweden.
Tel.: +34 924 289412; fax: +34 924 289412;
e-mail: amarzal@unex.es


Avian malaria parasites are supposed to exert negative effects on host fitness because these intracellular parasites affect host metabolism. Recent advances in molecular genotyping and microscopy have revealed that coinfections with multiple parasites are frequent in bird–malaria parasite systems. However, studies of the fitness consequences of such double infections are scarce and inconclusive. We tested if the infection with two malaria parasite lineages has more negative effects than single infection using 6 years of data from a natural population of house martins. Survival was negatively affected by both types of infections. We found an additive cost from single to double infection in body condition, but not in reproductive parameters (double-infected had higher reproductive success). These results demonstrate that malaria infections decrease survival, but also have different consequences on the breeding performance of single- and double-infected wild birds.


Two million people are killed every year by malaria. This death toll of human malaria is accompanied by another half billion people infected with Plasmodium, the causative agent of the illness (Teklehaimanot & Singer, 2005). But the systematic and ecological diversity of malaria parasites is much larger. For instance, more than 200 parasite species of the genera Plasmodium, Haemoproteus and Leucocytozoon have been described among the 4000 bird species investigated worldwide (Bishop & Bennett, 1992; Valkiūnas, 2005). This group of parasites is widespread, abundant and diverse and are easily sampled without disrupting the host populations. All these characteristics turn bird blood parasites into an excellent model for the study of host–parasite interactions.

Until relatively recently, avian haemosporidians were mainly considered harmless (Weatherhead & Bennett, 1992; Davidar & Morton, 1993; Dufva & Allander, 1995). However, when invading naive populations, these parasites can result in enormous mortality rates (Van Riper et al., 1986; Atkinson et al., 2000; Valkiūnas, 2005). Recent studies have found that these parasites can increase the probability of the host being killed by predators (Navarro et al., 2004; Møller & Nielsen, 2007) and they can have important effects on life-history traits in natural populations of their avian hosts, by affecting survival (Dawson & Bortolotti, 2000; Valkiūnas, 2005), body condition (Valkiūnas et al., 2006) and reproductive success (Merino et al., 2000; Marzal et al., 2005; Tomás et al., 2007).

Parasitized individuals may frequently carry several different parasites. Studies based on microscopic examination of blood smears have shown that mixed infections are common in many bird–parasite systems (Valkiūnas et al., 2003; Palinauskas et al., 2005). In addition, recent advances in methods of genotyping have shown that the number of avian malaria species is much higher than can be distinguished by traditional methods (Bensch et al., 2004), revealing coinfections by different parasite lineages or genera (Hellgren, 2005; Pérez-Tris & Bensch, 2005).

Multiple infections could be especially injurious for hosts, leading to anaemia, loss of body mass and reduced survival (Graham et al., 2005; Davidar & Morton, 2006). Such effects of malaria double infections have been poorly investigated and the reported results are inconclusive. For instance, Sanz et al. (2001) found no relationship between primary reproductive parameters and the number of blood parasite species infecting female pied flycatchers Ficedula hypoleuca. On the other hand, Evans & Otter (1998) showed a lethal combined effect of infections with Haemoproteus and Leucocytozoon in juvenile snowy owls (Nyctea scandiaca), although both parasite species on their own were not considered to be pathogenic.

In this study, we investigated the effects of single and double malaria infections on survival, body condition and reproductive success of their host, the house martin Delichon urbica, using the performance of uninfected individuals as a reference. This species is particularly suitable for this kind of study because it is colonial and hence suffers high parasite pressure (de Lope et al., 1993; Christe et al., 1998). The high breeding densities in colony breeding birds promote the transmission rate of parasites and frequently results in high levels of prevalence (Tella, 2002). The presence of different malaria parasite lineages infecting house martins was identified using molecular methods. If the infection with multiple parasite lineages is more costly than the infection of one parasite, then we would predict a negative trend in survival, body condition and reproductive success from uninfected to single-infected to double-infected house martins. To be precise, we would expect that double-infected individuals should survive less well to the next breeding season than healthy or single-infected ones. Moreover, double-infected individuals should be in lower body condition (e.g. lower body mass) than single-infected and healthy individuals and suffer a higher risk of being infected with other kinds of organisms, for example, ectoparasites. Finally, double-infected individuals should also show a reduction in reproductive parameters (e.g. laying date, clutch size and number of fledgings).

Materials and methods

Study site and collecting samples

The study was carried out in a colony of house martins in the surroundings of Badajoz (38°50′N, 6°59′W), south-west Spain, during a 6-year period (2002–2007).

From February to July, we followed reproductive events and collected blood samples from breeding individuals. Nests were inspected every second day to determine the start of laying. When the nestlings were 9 days old, we captured adult house martins at dawn in their nests and recorded tarsus length with a digital calliper to the nearest 0.01 mm, and body mass with a Pesola spring balance to the nearest 0.5 g. Feathers were inspected for the presence of chewing lice and feather mites by holding the extended wing and tail against the light and counting parasites (see Christe et al., 2002 for detailed information). We took a blood sample in a capillary tube for measurement of haematocrit. After centrifugation, blood cells were stored in 500 μL of 96% ethanol at room temperature.

We performed more than 30–40 capture sessions, each lasting several hours, in each population every year. This enabled us to build a capture history for each individual, indicating if the bird was encountered each season, and, if not, whether it was known to be alive as reflected by capture in a later year. The observation that 90% of house martins captured during the last capture sessions had already been captured previously during the same breeding season strongly suggests that very few birds escaped capture. In our study area, house martins show high breeding site fidelity, because they have never been recorded in another site in year (+ 1) after having bred in year i in our colonies. In addition, only 2.5% of house martins (= 3818) captured in a year were absent the next season but recaptured in following years. Therefore, we assumed that birds that were captured in a year and were not recaptured in subsequent years had not survived, as performed in other studies (Saino et al., 1997; Brown & Brown, 1999; Dawson & Bortolotti, 2000; Hanssen, 2006).

Molecular detection of blood parasite infections

Parasites were detected from blood samples using recently developed molecular methods (Bensch et al., 2000; Hellgren et al., 2004; Waldenström et al., 2004). DNA from the avian blood samples were extracted in the laboratory using standard chloroform/isoamylalcohol method (Sambrook et al., 2002). Diluted genomic DNA (25 ng μL−1) was used as a template in a polymerase chain reaction (PCR) assay for detection of the parasites using nested-PCR protocols described by Waldenström et al. (2004). The amplification was evaluated by running 2.5 μL of the final PCR on a 2% agarose gel. All PCR experiments contained one negative control for every eight samples. In the very few cases of negative controls showing signs of amplification (never more than faint bands in agarose gels), the whole PCR-batch was run again to make sure that all positives were true. Parasites detected by a positive amplification were sequenced using the procedures described by Bensch et al. (2000). Amplified fragments were sequenced from 5′-end with HaemF. The obtained sequences of 478 bp of the cytochrome b were edited, aligned and compared in a sequence identity matrix using the program BioEdit (Hall, 1999). Mixed infections were identified as ‘double base calling’ in the electropherogram. Parasites with sequences differing by one nucleotide substitution were considered to represent evolutionary independent lineages (Bensch et al., 2004; Ricklefs et al., 2005).

Statistical procedures

Logistic regression analysis was used to explore whether parasite infection status (double infected, single infected and uninfected), age (in years), year, sex, body mass, abundance of chewing lice and haematocrit influenced survival probability. Survival was treated as a binary variable (survivor vs. nonsurvivor). Parasite infection status was treated as a categorical variable in this analysis.

We used generalized linear models (GLM) to investigate the effect of age, sex, year (i.e. environmental variation) and infection status (uninfected, single or double infected) on three individual breeding traits (laying date, clutch size and brood size) and on three individual measures of body condition (body mass, haematocrit value and the abundance of chewing lice on the tail feather). We controlled for the effect of laying date on clutch size by including laying date in the model. Laying date and brood size were also included as predictors in GLM when haematocrit was the response variable, because it has been shown in previous studies that these reproductive traits could affect the haematocrit value (Gessaman et al., 1986; Hõrak et al., 1998). We used a normal error distribution with an identity link function to explain total variation in laying date (square root transformed), body mass and haematocrit value and a Poisson error distribution with a log link function to model clutch size, brood size and chewing lice (McCullagh & Nelder, 1989). There was missing information for some individuals, which resulted in slightly varying sample sizes in different analyses. The statistical significance of each covariate was tested in turn using a backward stepwise procedure. Models with different independent variables were compared using F-tests (Crawley, 2002). The final model only included significant explanatory variables. We used the GLM procedure of S-Plus 2000 and significant level was set at < 0.05 (Mathsoft, 1999).


Blood parasites and survival

We analysed 112 blood samples from house martins in search for blood parasites. We found 32 (29%) uninfected individuals and 80 (71%) individuals infected with blood parasites. Among the infected birds, 62 (55%) were infected by only one parasite lineage, whereas 18 (16%) carried a double infection. We found nine different blood parasite lineages, of which three were of the genus Haemoproteus (Durb1, Durb2 and Durb3) and six of the genus Plasmodium (SGS1, GRW2, GRW4, GRW11, Durb4 and Durb5) (Table 1).

Table 1.   Cytochrome b lineages and tentative parasite species for the house martin samples analysed.
Cytochrome b lineageParasite speciesGenbank accession no.N
  1. Number of host infected per each parasite lineage found and GenBank accession numbers are shown.

Durb1Haemoproteus sp.EU15434335
Durb2Haemoproteus sp.EU15434438
Durb3Haemoproteus sp.EU1543451
Durb4Plasmodium sp.EU1543461
Durb5Plasmodium sp.EU1543471
SGS1Plasmodium relictumAY56037216
GRW2Plasmodium ashfordiAY5603734
GRW4Plasmodium relictumAY0990411
GRW11Plasmodium relictumAY8317481

We analysed survival in relation to blood parasite infection (uninfected, single infected and double infected), the year when the sample was taken, the sex and age of individuals, body mass, abundance of chewing lice and haematocrit. Only infection status significantly explained variation in survival (Table 2). Double-infected individuals had the lowest chance of survival [percentage (SE) 17 (8.9)], single-infected individuals intermediate [23 (5.3)%] and healthy individuals the highest [47 (9.0)%]. There were no differences in survival probability between double- and single-infected house martins (Pearson χ2 = 0.29, d.f. = 1, = 0.589).

Table 2.   Results of a logistic regression analysis of factors determining the probability of survival of house martins.
  1. Infection statuses (1) and (2) represent the dummy variables which code for infection status in the equation.

Infection status  6.16320.046 
Infection status (1)1.2450.8272.26710.1323.475
Infection status (2)0.0440.8220.00310.9581.045
Body mass0.0100.1530.00410.9501.010
Abundance of chewing lice−0.0350.232.32610.1270.966

Infection status and reproduction investment

Double-infected and uninfected birds initiated clutches earlier than single-infected ones. Moreover, double-infected birds laid larger clutches compared with single-infected or uninfected house martins. Only infection status explained significant variation in laying date and clutch size (Table 3), and there was no effect of age, sex or year. Laying date of double-infected individuals [mean (SD) 110.5 (14.3) days] was similar to that of uninfected birds [108.1 (15.7) days] and significantly earlier than that of single-infected individuals [125.2 (17.9) days] (Fig. 1a). In addition, clutch size of individuals with double infection was significantly larger than those of single-infected and uninfected house martins [uninfected: mean (SD) 4.23 (0.82) eggs; single infected: 4.09 (0.95) eggs; double infected: 4.83 (0.79) eggs] (Fig. 1b). In addition, we found that infection status influenced number of fledglings while controlling for other confounding variables such as age, sex and year [uninfected: 3.60 (1.00) chicks; single infected: 3.26 (1.32) chicks; double infected: 3.94 (1.30) chicks] (Fig. 1c). Among the infected birds, double-infected individuals produced a significantly higher number of fledglings than single-infected ones (anova, F1,78 = 4.548; < 0.05).

Table 3.   Reduced GLM for laying date, clutch size and number of fledging for male and female individual house martins as response variables and haematozoan infection status, year, sex and age as predictor variables.
Variable Estimate (SE)Resid d.f.FP
  1. Laying date was also included in GLM with clutch size as response variable. Only significant terms are shown.

Lay date
 Null  104  
 Infection status 10.394(0.085)   
 Infection status 2−0.091(0.0065)10212.760.0001
Clutch size
 Null  104  
 Infection status 10.0013(0.060)   
 Infection status 20.0342(0.040)1024.630.011
 Laying date−0.003(0.002)1018.970.004
Number of fledging
 Null  104  
 Infection status 1−0.049(0.060)   
 Infection status 20.047(0.044)1022.280.10
Figure 1.

 Reproductive parameters for uninfected (= 18), single-infected (= 62) and double-infected house martins (= 32) of (a) laying date, (b) clutch size and (c) number of fledglings. Laying date is the number of days until laying starts (day 1 is 1 January). Error bar plots show mean ± 95% of confidence interval.

Body condition and infection status

The three estimates of body condition – body mass, abundance of chewing lice and haematocrit levels – varied with infection status but not with age, sex or year (Table 4). Individuals infected with two parasites were in poorer condition than the other birds. Specifically, double-infected birds had lower body mass than single-infected or uninfected individuals [uninfected: 17.06 (1.64) g; single infected: 16.76 (1.59) g; double infected: 15.83 (1.29) g] (Fig. 2a). Furthermore, individuals with double infection also suffered higher chewing lice intensities [uninfected: 10.19 (13.19); single infected: 7.92 (10.39); double infected: 20.06 (21.07)] (Fig. 2b) and double-infected house martins showed the highest haematocrit levels [uninfected: 54.13 (2.97)%; single infected: 52.65 (3.74)%; double infected: 55.78 (2.53)%] (Fig. 2c).

Table 4.   Reduced GLM with body mass, number of chewing lice and haematocrit value for male and female individual house martin as response variables and haematozoan infection status (as a factor).
VariableEstimate (SE)Resid. d.f.FP
  1. Year, age and sex were included in a maximal model for all response variables and, laying date and brood size were also included in a full GLM when haematocrit was the response variable. Only significant terms are shown.

Body mass
 Null 107  
 Infection status−0.152 (0.170)   
 Infection status−0.359 (0.135)1053.680.02
Chewing lice
 Null 107  
 Infection status−0.126 (0.035)   
 Infection status0.267 (0.021)1055.180.007
 Null 103  
 Infection status−0.737 (0.367)   
 Infection status0.796 (0.290)1016.570.002
Figure 2.

 Body condition parameters for uninfected (= 18), single-infected (= 62) and double-infected house martins (= 32) of (a) body mass, (b) abundance of chewing lice and (c) haematocrit (%). Abundance of chewing lice is the number of lices in tail feathers, and haematocrit represents the relative volume of red blood cells compared with total blood volume. Error bar plots show mean ± 95% of confidence interval.


Our findings show that the infection with malaria parasites had detrimental effects on house martins, as shown by the decreased survival prospects of double- and single-infected individuals. The negative effects of infection were detectable in reduced body condition of double-infected house martins, but not in single-infected ones. Contrary to our predictions, individuals harbouring a double infection invested more in current reproduction, despite being in poor physical condition. Next, we discuss these main results in detail.

A total of 18 house martins were recorded to carry double infections. From these, 15 individuals were infected with parasites belonging to the genus Haemoproteus (Durb1–Durb2), whereas three birds were mixed infected with different genera, Plasmodium and Haemoproteus (SGS1–Durb1 and SGS1–Durb2). These differences could be interpreted as a consequence of the variation in prevalence of each parasite lineage on house martin population. Moreover, differences in life cycles and vectors for parasites of the two genera may contribute to more frequent intrageneric mixed infections because mixed infections with parasites of two different genera require bites from multiple vectors (Valkiūnas, 2005). Unfortunately, the low number of PlasmodiumHaemoproteus double infections found in our population makes it impossible to evaluate whether double infections with parasites from one or two genera are associated with different levels of fitness costs.

Parasites, by definition, have detrimental effects on their hosts (Noble & Noble, 1976; Price, 1980). Although haematozoan-induced mortality is frequently reported in domestic birds (Atkinson & van Riper, 1991; Valkiūnas, 2005), most studies of wild birds have failed to document such effects (Davidar & Morton, 1993; Shutler et al., 1996; Schrader et al., 2003; Bensch et al., 2007, but see Sorci & Møller, 1997; Dawson & Bortolotti, 2000). In a previous study, we found that the intensity of Haemoproteus infections negatively affected survival in yearling house martins, suggesting that intensity of parasitism was a significant predictor of survival in this population (A. Marzal, M. Reviriego, J. Balbontin, C. Relinque, F. de Lope and A.P. Møller, unpublished data). Infections of hosts by multiple strains of the same parasitic species occur in many host–parasite systems (Read & Taylor, 2001), which may select for higher parasite virulence (Mosquera & Adler, 1998; Taylor et al., 1998; but also see Taylor et al., 2002; de Roode et al., 2003). In addition to this, host may have difficulties to mount an immune response against more than one parasite genotype (Combes, 2001). In the present study, only 17% of double-infected individuals survived, compared with 23% of single-infected and 47% of uninfected individuals, showing the lethal effects of malaria infection.

In agreement with our predictions, double-infected birds were in a poor body condition, having the lowest body mass and the highest infestation with ectoparasites. However, these negative effects were not evident on single-infected individuals. Blood parasites may have detrimental effects on body mass (Dufva, 1996; Merino et al., 2000; but also see Bennett et al., 1988) and consequently may reduce adult bird survival by increasing starvation risk (Cuthill & Houston, 1997) or reducing the immune response (Møller et al., 1998). On the other hand, double-infected birds also harboured the higher intensity of ectoparasites. Because avian preening is a time and energy demanding activity (Giorgi et al., 2001; Moore, 2002), the highest levels of ectoparasites found in double-infected birds could be interpreted as a consequence of the weakness caused by malaria infection.

The use of the haematocrit levels as indicators of condition of wild birds has recently been disputed (see review in Fair et al., 2007). Several studies have reported decreased levels in infected birds (Møller, 1991; Richner et al., 1993; Moreno et al., 1999; Potti et al., 1999; Simon et al., 2005), whereas others have reported no change (Shutler et al., 1996; Ots & Hõrak, 1998; Valera et al., 2005) or even increased levels (Dawson & Bortolotti, 1997; Johnsen & Zuk, 1998; Merino et al., 2001). Contrary to our predictions, we found that double-infected birds showed the highest levels of haematocrit, whereas single-infected house martins had the lowest levels. Hence, it appears that the haematocrit levels of house martins in this study are not a straightforward indicator of body condition when assessing the pathogenicity of blood parasites. An explanation to the observed pattern could be that the higher haematocrit levels in parents of larger broods (double-infected house martins) could be an adaptive physiological response to the higher requirement of efficient oxygen transport when rearing more young, as some other studies have found (Gessaman et al., 1986; Hõrak et al., 1998).

Previous experimental studies on passerine birds reported detrimental effects of haematozoan parasites on breeding performance, negatively affecting clutch size and breeding success (Merino et al., 2000; Marzal et al., 2005). In the present study, we show that the infection with one parasite lineage has negative consequences on the fitness of house martins. We expected to observe a trend in pathogenic consequences from uninfected to single-infected to double-infected birds. Surprisingly however, double-infected and uninfected house martins laid their clutches earlier than single-infected ones, which can lead to higher reproductive success (Crick et al., 1993). In addition, clutches from double-infected birds were larger and produced more nestlings. These observations contradict the negative effect of double infections previously seen in survival and body condition. One explanation to reconcile these results could be that double-infected individuals are trying to maximize current reproduction when survival prospects are challenged. Following this line and according to the terminal investment hypothesis, residual reproductive value decreases with increasing age or risk of death due to parasites. Individuals with a lower chance of survival (e.g. old or sick individuals) are then expected to invest all their energy and effort in current reproduction (Williams, 1966; Clutton-Brock, 1984). Recently, several studies have shown that blood parasite infection and other diseases may initiate terminal investment behaviours (Sanz et al., 2001; Bonneaud et al., 2004; Hanssen, 2006; Velando et al., 2006). Our results show that both single and double infections result in a decrease in survival prospects. But only double-infected birds invested more in reproduction, and it remains to be explained why single-infected birds did not. Hosts may face up to parasite challenge in different ways to maximize their fitness. First, hosts that are likely to control or clear the infection should invest in immune response and maintenance (Sandland & Minchella, 2003) waiting for more favourable future breeding seasons, despite trade-offs between immune function and other important life-history traits such as condition, reproduction and survival (Boots & Bowers, 1999; Marzal et al., 2007). Alternatively, when challenged with an infection that is likely to be lethal, hosts may develop a different strategy termed ‘fecundity compensation’ or ‘terminal investment’ which increases fitness through early reproductive enhancement (Minchella & LoVerde, 1981; Sandland & Minchella, 2004). Experimental studies are needed to conclusively evaluate the role of double and single infections in resource allocation strategies employed.

In conclusion, we documented several differences in fitness effects of single and double malaria infections on house martins. Besides clear effects of parasites on survival, our results showed mixed effects on body condition and reproductive success, suggesting that the simple rule of ‘one plus one is two’ is not applicable to this particular host–parasite system.


Carmen Relinque kindly helped in field works. Research was funded by research project of the Spanish Ministry of Education and Science (CGL2006-2937) to FdL and AM and the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (FORMAS) to SB. AM, MIR and JB were supported by post-doctoral (EX-2006-0557), predoctoral (BES 2004-4886) and post-doctoral programme ‘Juan de la Cierva’ grants from Spanish Ministry of Education and Science respectively. We are grateful to two anonymous reviewers for suggestions to improve the article. All the experiments comply with the current laws of Spain, where the experiments were performed.