North-African house martins endure greater haemosporidian infection than their European counterparts



Afro-Palearctic migrant species are exposed to parasites at both breeding and over-wintering grounds. The house martin Delichon urbicum is one such migratory species facing high instances of blood parasite infection. In an attempt to determine whether breeding European house martins harbour similar blood parasite communities to populations breeding in North Africa, birds were sampled at their breeding grounds in Switzerland and Algeria. Moreover, haemosporidian prevalence and parasite communities were compared to published data sets on Spanish and Dutch breeding populations. This study furthermore wanted to establish whether co-infection with multiple genera or lineages of parasites had negative effects on host body condition. Breeding house martins caught in Algeria showed a higher prevalence of avian haemosporidian parasites than did European populations. Swiss house martins showed a prevalence comparable to that of Spanish and Dutch populations. There were slight differences in the haemosporidian community between European and North-African populations in terms of composition and abundance of each lineage. Similar to the Dutch house martins, but in contrast to the Spanish population, infection status and number of genera of parasites infecting single hosts did not influence Swiss house martin body condition.

Bird migration is one of the most well-known global phenomena that has captivated and enthralled scientists and non-scientists alike by the birds’ excellent homing abilities and their amazing physiological feats of endurance (Zink 2011). Migrating birds have to optimize their migration strategy in terms of timing, stop-overs and energy acquisition (Schmaljohann et al. 2012). An additional challenge facing migrating species can be exposure to parasites at both their breeding and over-wintering grounds, thereby exposing them to a broader range of parasites (Waldenström et al. 2002, Hellgren et al. 2007b, Jenkins et al. 2012) and indeed, parasitism can add to the energetic costs of migration and be related to delayed arrival times at the breeding grounds (Møller et al. 2004).

Resident African birds might act as reservoirs of parasites which could infect Afro-Palearctic migrants upon their arrival, and this in turn could result in the movement of parasites across continents. However, in a recent study on blackcaps Sylvia atricapilla, Santiago-Alarcon et al. (2011) investigated whether populations following different migration routes with separate overwintering grounds had divergent haemosporidian parasite communities, but found no such difference. In contrast, over an evolutionary timescale Hellgren et al. (2007b) found little evidence for a shift in transmission areas of haemosporidian parasites; there were distinct genetic signatures of African and European transmitted parasites. Therefore, more research on whether migrants act to homogenise parasite communities or not is needed.

Avian haemosporidian blood parasites are a group of vector-transmitted parasites belonging to three genera: Plasmodium, Haemoproteus and Leucocytozoon. So far, studies have shown that they can contribute to decreased survival in American kestrels, Falco sparverius (Dawson and Bortolotti 2000, Valkiūnas 2005), increased predation in a broad range of species (Navarro et al. 2004, Møller and Nielsen 2007) and lower reproductive success in breeding populations of both the house martin Delichon urbicum (Marzal et al. 2005) and the blue tit Cyanistes caeruleus (Merino et al. 2000, Marzal et al. 2005, Tomás et al. 2007). In migratory blackcaps, infection coincides with slightly later spring arrival (Santiago-Alarcon et al. 2013). On the other hand, a study on the purple martin Progne subis has shown no effect of infection on survival and return rates and even a positive effect on fledging success (Davidar and Morton 1993). Furthermore, co-infection, concurrent infection with multiple parasite lineages can also lead to a higher reproductive success in some cases (Marzal et al. 2008). Therefore, the effects of infection on haemosporidians can be idiosyncratic and are dependent on the host- parasite combination under study (Rätti et al. 1993).

The house martin breeds in colonies of ten to hundreds of pairs. Due in part to their coloniality, house martins suffer a high level of parasitism by many species, in particular from around 12 flea species (Tripet et al. 2002, Pilgrim and Galloway 2003, Møller et al. 2005), the house martin bugs (Oeciacus hirundinis, Hemiptera: Cimicidae and the louse fly, Stenepteryx hirundinis, Diptera: Hippoboscidae (de Lope et al. 1993, Christe et al. 2000, 2001). They are also infected by haemosporidians, and recent studies by Marzal et al. (2008) and Piersma and van der Velde (2012) have investigated haemosporidian community composition in breeding colonies of house martins in Spain and the Netherlands. These studies further document how body mass and reproductive success is affected by infection with one or more parasite lineages, or haplotypes, (see Christe et al. 2002 for the effect of Haemoproteus prognei in a Spanish colony on different breeding and physiological parameters). These two house martin populations had four lineages in common and apart from a few occurrences of other lineages, each had one lineage present in substantial numbers that was not detected in the other population (lineage SGS1 for Spain and GRW9 for the Netherlands). Infection status proved to be an important determinant of body condition in the Spanish population (Marzal et al. 2005, 2008) but no such effect was present in the Dutch study (Piersma and van der Velde 2012).

For the purposes of this study, we sampled house martins from a European and an African location in order to determine whether breeding European populations’ haemosporidian communities are more similar to those on the African continent in the southernmost limits of their breeding range. Moreover, we assessed the Leucocytozoon prevalence and lineage identities in house martins, information lacking in previous studies on this species. We furthermore wanted to assess whether infection and co-infection (i.e. concurrent infection either with different genera or lineages of parasite) with haemosporidians are correlated to negative effects in terms of body condition of Swiss breeding house martins.


Bird sampling

Adult house martins were caught on their nests at the end of the reproductive period in 2011 in Annaba, Algeria (36°54’N, 7°45’E) and in Conthey, Switzerland (46°14’N, 7°18’E). A blood sample was taken by puncturing the brachial vein with a sterile 25 gauge needle. Blood from Algerian house martins was then collected with capillary tubes and a drop was deposited on standard filter paper. For Swiss house martins the resulting blood drop was collected in a lithium-heparin lined Microvette.

Molecular analyses

DNA extraction from blood was done using the DNeasy blood and tissue extraction kit according to the manufacturer's protocol for purification of DNA from blood using the BioSprint 96. Following DNA extraction, a nested PCR, amplifying a 479 bp region of the parasite cytochrome b gene (Hellgren et al. 2004, modified in van Rooyen et al. 2013) was performed on all samples.

Nested PCR products were purified using the Wizard SV Gel and PCR Clean-Up System using the manufacturer's protocol for DNA purification by centrifugation. Purified PCR products were sequenced both in the forward and reverse directions using the primers HaemF and HaemR2 (Plasmodium/Haemoproteus) or HaemFL and HaemR2L (Leucocytozoon), identified by performing a local BLAST search against the MalAvi database (Bensch et al. 2009).

Analysis of co-infections

Co-infections were categorised in two ways: 1) co-infection with multiple lineages of different genera: Leucocytozoon and Plasmodium/Haemoproteus if the sample amplified with both primer pairs. 2) Co-infection with multiple lineages from the same genus: by visual inspection of the chromatogram, apparent as double-base callings. This latter method however, does not fully resolve lineages in co-infections or discriminate between Plasmodium–Haemoproteus, Plasmodium–Plasmodium or Haemoproteus–Haemoproteus co-infections, which are amplified in the same PCR reaction. The only way to fully discriminate among such co-infections is to clone, however this was beyond the budget of this project.

Phylogenetic reconstruction

A Bayesian phylogeny of the blood parasites that have been found in house martins to date was built, using BEAST ver. 1.7.5 (Drummond and Rambaut 2007), and applying a relaxed molecular clock with Yule speciation prior. Firstly, we tested for the most likely model of evolution using the online server FindModel: (<>). The GTR +Γ among site variation model had the lowest AIC and so we used this as a prior. The mean substitution rate was fixed to 1.0, thereby fixing the rate of substitution of internal nodes to substitutions/site. A first chain was run to check for convergence, for 10 million generations, sampling every 1000 generations. After visual inspection of the trace file in Tracer ver. 1.5 (Drummond and Rambaut 2007) it appeared that the run had not yet converged, and so the three further runs were re-run for longer (30 million generations, sampling every 1000 generations).

Statistical analyses

All analyses were performed with the freeware R-Cran Project, ver. 2.15.2 (<>). Prevalence and lineage information that we obtained were combined with previously published data sets of house martin blood parasites (Marzal et al. 2008, Piersma and van der Velde 2012). Prevalences were calculated with 95% confidence intervals using the program Quantitative Parasitology 3.0 (Rόzsa 2000). To test whether there were differences in the overall infection prevalence, the prevalence of each genus (Plasmodium, Haemoproteus and Leucocytozoon) and of the main lineages – and where data were available for morphospecies – we compared sites using χ2 or Fisher's exact tests (for situations when counts were fewer than five individuals per site). The prevalence of Leucocytozoon spp. was only compared between the Swiss and Algerian populations, as these data were not available for the other sites.

For thirty house martins caught at the Swiss site, we obtained body mass and morphometric data such as wing length, measured to the nearest mm, and tarsus length, measured using digital callipers to the nearest 0.01 mm. We then investigated whether body mass or a measure of condition were related to co-infection status (uninfected, single infection or co-infection, either with multiple genera or lineages), or to infection with a specific genus (Plasmodium, Haemoproteus or Leucocytozoon). These two variables were analysed separately. To measure condition, we wanted to use the scaled mass index (Peig and Green 2009), which accounts for the allometric scaling of body measurements. However, as the slope of the ordinary least squares regression of ln(mass) on ln(tarsus)/ln(wing) was not significant, which is a prerequisite for calculating the scaling exponent (Legendre and Legendre 1998), it was thereby not possible to calculate the scaled mass index, and conduct this analysis. As an alternative measure of condition, we used mass over wing length. Wings perform a key functional role in migratory species, and therefore could be an important component of condition (Merom et al. 2000). We conducted Shapiro tests to assess the normality of body mass and arcsine square root transformed mass over wing, as the values for the latter were between zero and one. The Shapiro test rejected a normal distribution for mass and all other data transformations of mass, but confirmed normality for the transformed mass over wing variable. Transformed mass over wing was analysed using a one-way ANOVA with either co-infection status or genus infection as the explanatory variable. Body mass was analysed using the non-parametric equivalent, i.e. the Kruskal–Wallis test.


Haemosporidian infection in house martins

A total of 30 adult house martins were captured in the Algerian and 43 in the Swiss population. We found a significant difference in the overall prevalence of blood parasites among all four populations, including Spain and the Netherlands (χ2= 8.66, DF = 3, p = 0.03, Table 1). The Algerian population showed the highest prevalence, and Swiss house martins showed a 79% prevalence for Plasmodium and Haemoproteus which is similar to the 71% prevalence found in the Spanish and 77% in the Dutch populations, respectively.

Table 1. The prevalence and 95% CI of avian haemosporidian blood parasites in four populations of house martin from Spain (Marzal et al. 2008), the Netherlands (Piersma and van der Velde 2012), Algeria and Switzerland (our data). *Out of the five Algerian co-infected individuals, two were co-infections with Leucocytozoon and three were co-infections involving Plasmodium, Haemoproteus, yet we were not able to disentangle the two in our sequencing protocol. Out of the six Swiss co-infected individuals, three were co-infections with Leucocytozoon and another three were co-infections involving Plasmodium, Haemoproteus, though were not able to disentangle the two in our sequencing protocol. ‘−‘ indicates missing data, or data that could not be inferred from previously published work
 AlgeriaSpainSwitzerlandThe Netherlands
 n (ncoinf)prev (95% CI )n (ncoinf)prev (95% CI )n (ncoinf)prev (95% CI )n (ncoinf)prev (95% CI )
Overall infection29 (5*)0. 97 [0.82–0.99]80 (18)0.71 [0.62–0.79]34 (6)0.79 [0.64–0.89]276 (151)0.77 [0.72–0.81]
Haemoproteus spp.20 (–)0.67 [0.48–0.83]74 (–)0.66 [0.57–0.75]31(–)0.51 [0.36–0.66]247 (139)0.69 [0.64–0.74]
Plasmodium spp.6 (–)0.2 [0.09–0.38]24 (–)0.21 [0.15–0.30]10 (–)0.23 [0.12–0.38]29 (12)0.08 [0.06–0.11]
Leucocytozoon spp.2 (2)0.07 [0.01–0.21]4 (3)0.09 [0.03–0.22 ]

Twelve lineages were observed in our two study sites out of 17 lineages found across the four populations sampled so far (Table 2, Fig. 2). Out of these, three (DELURB1, DELURB2, GRW2) were also present in the Spanish and Dutch samples. SGS1 was also a lineage in common to our two sampling locations, as well as to Spain. GRW11 found in the Algerian samples, had also been previously detected in Spain and the Netherlands. GRW4 on the other hand, was found in both Spanish and Algerian birds. HIRUS7, a lineage of Leucocytozoon, was found in both Algerian and Swiss house martins. Five lineages (DELURB7, GRW9, ANLAT14, PARUS19 and DELURB9) were only detected in the Swiss population, of which two were newly discovered (DELURB7 and DELURB9, GenBank accession no.: JX867109, JX867110).

Table 2. The number of individuals infected by each lineage (n), with associated prevalence (95% CI) in four populations of house martin sampled in Algeria and Switzerland by us and in Spain (Marzal et al. 2008) and the Netherlands (Piersma and van der Velde 2012); *new lineages, – no information available. Individuals suffering multiple infections for the Swiss and Algerian populations are excluded
 GenBank accession no. AlgeriaSpainSwitzerlandthe Netherlands
LineageMorphospeciesnprev (95% CI)nprev (95% CI)nprev (95% CI)nprev (95% CI)
DELURB1EU154343 Haemoproteus spp.60.2 [0.09–0.38]350.31 [0.23–0.41]130.3 [0.18–0.45]2030.57 [0.51–0.62]
DELURB2EU154344 Haemoproteus spp.140.47 [0.30–0.65]380.34 [0.25–0.43]80.19 [0.09–0.34]1100.31 [0.26–0.36]
DELURB3EU154345 Haemoproteus spp.010.01 [0.0005–0.05]00
DELURB7*JX867109 Haemoproteus spp.0010.02 [0.001–0.12]0
DELURB4EU154346 Plasmodium spp.010.01 [0.0005–0.05]00
DELURB5EU154347 Plasmodium spp.010.01 [0.0005–0.05]00
SGS1AF495571 Plasmodium relictum 30.10 [0.03–0.26]160.86 [0.78–0.91]30.07 [0.02–0.19]0
GRW2AF254962 Plasmodium ashfordi 10.03 [0.002–0.18]40.04 [0.012–0.088]50.12 [0.05–0.25]130.036 [0.02–0.061]
GRW9DQ060773 Plasmodium spp.0020.05 [0.008–0.16]140.04 [0.023–0.065]
GRW4AF254975 Plasmodium relictum 10.03 [0.002–0.18]10.01 [0.0005–0.05]00
GRW11AY831748 Plasmodium relictum 10.03 [0.002–0.18]10.01 [0.0005–0.05]020.006 [0.001–0.02]
COLL6DQ368375 Plasmodium spp.00040.011 [0.004–0.03]
TURDUS1HQ537478 Plasmodium circumflexum 00010.003 [0.0002–0.02]
HIRUS07JN164707 Leucocytozoon spp.20.07 [0.01–0.21]10.02 [0.001–0.12]
ANLAT14FJ839449 Leucocytozoon spp.010.02 [0.001–0.12]
PARUS19 Leucocytozoon spp.010.02 [0.001–0.12]
DELURB9*JX867110 Leucocytozoon spp.010.02 [0.001–0.12]

Within each genus, there was a significant difference in the prevalence of the two main lineages (Haemoproteus: DELURB1–DELURB2, χ2= 15.04, DF = 3, p = 0.002, Fig. 1) and Plasmodium relictum-ashfordi (Fisher's exact test p < 0.001, Fig. 1). Following the south–north cline from Algeria to Switzerland, the prevalence of DELURB1 markedly increased more than two-fold, and became the dominant parasite in the Dutch population. The Swiss population, on the other hand, appeared to have a higher proportion of GRW2 (P. ashfordi), than of P. relictum haplotypes. Data on Leucocytozoon infection was only available for our two study populations and we found no difference in the prevalence of Leucocytozoon infection between the two sites (Fisher's exact test: p = 1).

Figure 1.

A stacked bar plot showing the relative abundance of the four main lineages/morphospecies of haemosporidians in house martins in our data (Algeria and Switzerland) and data from the Spanish (Marzal et al. 2008) and Dutch populations (Piersma and van der Velde 2012). As there were several members of the P. relictum species complex and the overall prevalence is low, we pooled these individuals.

Figure 2.

The maximum clade credibility tree of all house martin lineages sampled to date based on three separate runs in BEAST ver. 1.5.7. A Hepatocystis lineage has been included to place the relationships in a broader phylogenetic context (note that this is not directly specified as an ‘outgroup’ in BEAST).

Frequencies of co-infection among the Plasmodium and Haemoproteus genera were 10.3% for the Algerian population and 5.9% for the Swiss one and when compared to the Spanish and Dutch house martin populations, there was no statistical difference in their frequency (χ2= 0.63, DF = 3, p = 0.89). Co-infection with Leucocytozoon sp. was 6.9% in Algeria and 11.8% in Switzerland.

Relationship between Haemosporidian infection and host body mass and body condition of the Swiss population

Body condition measured as mass over wing length was not related to any of the infection parameters, which included models fitted with either a term for genus identity or co-infection status (Table 3). Similarly, the results of the Kruskal–Wallis test rejected a link between mass and either co-infection status (KW χ2= 2.4, DF = 2, p = 0.30) or genus identity (KW χ2= 0.83, DF = 1, p = 0.36).

Table 3. ANOVA table of mass over wing length against a) co-infection status (uninfected, infected and co-infected either with multiple lineages or multiple genera) and b) genus (Plasmodium/Haemoproteus). It was not possible to account for Leucocytozoon spp. in this analysis, as they occurred only in co-infections
 DFSum of squaresMean squaresF-valuep
 Co-infection status20.0004090.0002050.9140.413


We were interested in whether European breeding house martin populations had a more similar haemosporidian parasite community composition than a population sampled in North Africa. We found that birds sampled from the Algerian population had a higher overall prevalence of infection than those sampled in Europe. Birds from each location harboured some rare lineages that were unique to each site, yet populations tended to share the main lineages, namely DELURB1, DELURB2, GRW2 (P. ashfordi) and members of the P. relictum species complex. Interestingly, the dominant parasite in the Algerian population was DELURB2, in contrast to DELURB1 in the northernmost population in the Netherlands. Data from the Swiss population showed that there was no association between body mass (or body condition) and infection status.

Significantly more house martins were infected in Algeria than in any of the European locations. The three genera of haemosporidians observed in this study are typically thought to be vectored by different blood-sucking dipterans. Haemoproteus is supposed to be transmitted by biting midges (Diptera: Ceratopogonidae), Plasmodium by mosquitoes from the family Culicidae and Leucocytozoon by black flies, Simuliidae (Valkiūnas 2005). It is worth noting however that detailed studies of vector competences and host preferences have lagged behind in the past thirty years. Therefore, the exact vector preferences and parasite associations may vary based on the structure of the co-evolutionary landscape (reviewed by Santiago-Alarcon et al. 2012). The higher prevalence observed in north Africa could be a by-product of the Mediterranean climate at Annaba which represents ideal conditions for vectors and could allow the spread and persistence of avian haemosporidians during the breeding season at this location. Future work should focus on vectors at this site. Leucocytozoon prevalence was generally low in the two populations that we sampled, and could also be related to lack of suitable vector oviposition sites and/or climatic variables (Crosskey 1990).

Despite having higher overall haemosporidian prevalence, Algerian birds did not differ much in their haemosporidian community from breeding populations in Europe. Algerian house martins shared the three dominant lineages (DELURB1, DELURB2 and GRW2) with all the European populations and a further one lineage with Spain only, one with Spain and the Netherlands and one more with Spain and Switzerland (Table 2). It therefore appears that the Algerian parasite community comprises a mixture of lineages from breeding populations around Europe. Interestingly, the dominant parasite lineage shifts from being DELURB2 in the Algerian population to DELURB1 in the Dutch one. However, whether this reflects a true north–south cline warrants further testing.

Comparable to the previous study on the Dutch population of house martins (Piersma and van der Velde 2012) we found no effect of infection or co-infection on house martin weight or body condition, measured as mass over wing length. In addition, we have to be cautious with this interpretation as only thirty individuals were included in this analysis. DELURB1 and DELURB2 are confined to hirundines and indeed, the fact that in two out of three populations studied, they do not appear to have any detrimental effects, indicates that perhaps this could contribute to the success of these parasites. Alternatively, these parasites could act on a different part of a host's life history that we have not measured. For instance, Marzal et al. (2008) found negative effects of infection and co-infection on house martin survival in their Spanish population, but positive effects of co-infection on host fitness, measured as clutch size. Taken together, these results further reveal the idiosyncratic nature of haemosporidian parasite–host interactions, and strongly supports why virulence and pathogenicity have to be determined on a population- by- population basis, even if it is the same host–parasite combination under study.

Only five lineages infecting house martins have been attributed to morphospecies (Bensch et al. 2009). These were SGS1, GRW4, GRW11 (P. relictum), GRW2 (P. ashfordi) and TURDUS1 (P. circumflexum). For the purposes of our analyses, and because the abundance of Plasmodium lineages was low, we combined all lineages belonging to the P. relictum morphospecies. However, though P. relictum occurs world-wide (Valkiūnas 2005), individual lineages from this morphospecies may indeed be quite different species. SGS1 is the most widespread of these parasites, and might be transmitted in eastern Europe (Beadell et al. 2006, Hellgren et al. 2007a), western, northern and southern European countries (Hellgren et al. 2007b, Glaizot et al. 2012), Nigeria (Waldenström et al. 2002) and South Korea (Beadell et al. 2006). GRW4 is a tropical parasite and associated with warm climate countries having been recorded in Hawaii, Nigeria and Papua New Guinea (Ricklefs and Fallon 2002, Waldenström et al. 2002, Beadell et al. 2006) but has not been recorded in resident birds in northern Europe (Valkiūnas et al. 2007). Although less is known about the transmission of GRW11, a lineage also within the P. relictum clade, it has been recorded in Europe in different countries (Dimitrov et al. 2010, Zehtindjiev et al. 2012). The life-cycle of P. ashfordi lineage GRW2 remains unknown but it is suspected that this lineage is only transmitted in Africa (Valkiūnas et al. 2007). The relatively higher prevalence in the Swiss house martin population is therefore worth noting. Transmission of P. circumflexum lineage, TURDUS1, occurs throughout Europe (Hellgren 2005, Hellgren et al. 2007a, Glaizot et al. 2012). As this lineage has not been recorded in Africa (Waldenström et al. 2002) it is considered to be a European transmitted lineage (Palinauskas et al. 2007).

The house martin is an African–Eurasian migrant. Even so, information on house martin migration routes in general is scarce and overwintering grounds are believed to be somewhere in sub-Saharan Africa (del Hoyo et al. 2004). A recent study by Evans et al. (2012) has contested the widely believed theory that house martins overwinter in moist montane forests (Morel and Morel 1992, Pearson and Lack 1992) by showing through stable isotope analysis that individuals of this species more likely spend their winter in dry, open habitats. For birds ringed in Switzerland, only two individuals have been recaptured to the south: one in St Aygulf, France and another in Tinerhir, Morocco (Maumary et al. 2007). There are no recaptures of Swiss house martins in the overwintering quarters, and therefore knowing where these birds get infected with their haemosporidian parasites remains to be investigated.


The results presented in this study might hint towards a mixture of breeding pools of house martins during migration. Different breeding populations most probably still choose different overwintering grounds as their blood parasite communities, believed to be transmitted mostly at the overwintering grounds (Bensch et al. 2007, Piersma and van der Velde 2012), are not similar enough to suggest a communal source. This does not exclude the possibility of co-migration however, for the benefit of increased protection against predation or increased locomotion efficiency (Andersson and Wallander 2004).


Juan van Rooyen and Tania Jenkins are co-first authors. This work was funded by the Swiss National Science Foundation (grants 31003A-120479 and 31003A-138187). We are grateful to Jessica Delhaye and François Biollaz for field assistance, to Raphaël Arlettaz for helping set up the nest boxes and to the staff of les ‘Fougères’ for giving us access to the study colony.