Geographic and host distribution of haemosporidian parasite lineages from birds of the family Turdidae

Species of the Turdidae Rafinesque, 1815, or thrushes, are small to medium-sized songbirds of the order Passeriformes, which inhabit all biogeographic realms. Most thrushes forage on the ground, where they feed on insects, earthworms, land snails, and fruit. Migration behavior can vary substantially between species, ranging from long-distance migrants to resident birds. The family currently includes about 160 species classified into 20 genera [11]. The most species-rich genera are Turdus (80 species), Geokichla (18), Zoothera (15), Catharus (12), Myadestes (12), Cochoa (4), and Sialia (3), while the other genera include only one or two species each. Of these genera, four are mentioned here because data were also included in the present study: Entomodestes (2), Neocossyphus (2), Hylocichla (1), and Ixoreus (1) [11]. Previously, birds of the family Muscicapidae were also classified as Turdidae, but they do not form a monophyletic group with them and are now considered an independent group [12].

So far, 14 haemosporidian parasite species have been described from bird hosts of the family Turdidae. These include five Plasmodium, four Haemoproteus, and four Leucocytozoon species (Table 1). One parasite species, Plasmodium lutzi was described from the Grey-necked wood rail Aramides cajaneus (Gruiformes) in Brazil, but is considered to be common in thrushes [13].

Thrushes were sampled for various studies on avian haemosporidians, but none of the molecular genetic studies particularly dealt with haemosporidian parasites of this host family. Most of the haemosporidian CytB sequences of thrushes come from a few ecological studies screening samples of large numbers of passeriform birds. [8] analysed CytB sequences recovered from more than 2300 birds from Western Europe, Western Russia, Western Asia, and Northern Africa, 186 of which originated from thrushes. [14] studied 69 bird communities from all over the Americas and published at least 86 sequences isolated from thrushes. Further data on haemosporidian parasites of Alaskan birds were published by [15, 16], featuring more than 100 sequences isolated from thrushes. Data from American thrushes were also published in ecological studies of [17‐19]. More data from haemosporidians of American thrushes were published in [14, 20‐25]. Additional sequence data originate from numerous other studies. All references for the CytB sequences used in this study are provided in Additional file 1.

Birds of the family Turdidae are among the most sampled host groups of avian groups, in which both the morphological and molecular diversity of haemosporidian parasites have been relatively well characterised in different zoogeographical regions [1, 26, 27]. This provided opportunities for the relatively representative comparative parasite taxonomic and distribution analysis and helped to determine unrecognised patterns in the distribution of avian haemosporidians. Among thrushes, Turdus merula is of particular interest because it was also introduced to European settlements in Australia and New Zealand together with its haemosporidian parasites, which potentially represent a threat to native bird species [28].

In the years 2003 to 2018, samples were collected from 310 individuals of T. merula, 36 individuals of T. philomelos, and two individuals of T. pilaris. Most samples (288) were taken from dead birds during a monitoring study at the Institute of Pathology (University of Veterinary Medicine Vienna) from 2003 to 2005. Another 15 samples were collected from dead birds between 2014 and 2017. After dissection, various organs were embedded in paraffin blocks and stored in the archive of the Institute of Pathology, whereas only native brain tissue was frozen and stored at − 80 °C for DNA isolation. Additional 43 blood samples were collected from living blackbirds and song thrushes received for treatment at the Bird and Reptile Clinic (Department for Companion Animals and Horses, University of Veterinary Medicine Vienna). Blood samples were taken by puncturing the brachial vein and using heparinised microcapillaries to transfer blood drops to high-grade filter papers Whatman™ 903 (GE Healthcare, Buckinghamshire, GB) of which DNA was isolated later.

From the tissue of dead birds, both DNA and RNA were isolated because these samples were originally used for the Usutu virus screenings [29‐32]. Nucleic acids were extracted from 140 µl of homogenised brain tissue with the QIAamp Viral RNA Mini Kit (QIAGEN, Hilden, Germany) as described in [32]. From blood spots, DNA was extracted using the DNeasy Blood & Tissue Kit (QIAGEN, Venlo, Netherlands) following the manufacturer’s protocol for isolation of DNA from tissue. In the last step, two eluates of 100 µl were made from the same column, at 8000 rpm and 13,000 rpm, the second of which was used for the PCR screenings.

All 348 samples were screened for the presence of avian haemosporidians using the nested PCR-protocol established by [4], which allows obtaining mt CytB fragments of 476 bp for Leucocytozoon species and 478 bp for Plasmodium spp. and Haemoproteus spp. In the “nest 1” PCR, the primers HaemNFI (5′-CATATATTAAGAGAANTATGGAG-3′) and HaemNR3 (5′-ATAGAAAGATAAGAAATACCATTC-3′) were used. In the “nest 2” PCRs, the primers HaemF (5′-ATGGTGCTTTCGATATATGCATG-3′) and HaemR2 (5′-GCATTATCTGGATGTGATAATGGT-3′) were used to amplify the CytB of Plasmodium spp. and Haemoproteus spp., and HaemFL (5′-ATGGTGTTTTAGATACTTACATT-3′) and HaemR2L (5′-CATTATCTGGATGAGATAATGGIGC-3′) were used to amplify the CytB of Leucocytozoon spp. The PCRs were performed using the GoTaq^® G2 Flexi DNA Polymerase (Promega, Wisconsin, Madison, USA). The PCRs started with an initial denaturation for 2 min at 94 °C, followed by 35 cycles with 30 s at 94 °C, 30 s at 50 °C, 1 min at 72 °C, and a final extension for 10 min at 72 °C. Each 1 µl of “nest 1” PCR-product was used as a template in the “nest 2” PCRs.

The PCR products were sent to Microsynth Austria GmbH (Vienna, Austria) for purification and sequencing in both directions using the “nest 2” PCR primers. Raw forward and reverse sequences were manually aligned and electropherograms were checked in Bioedit v.7.0.8.0 [33]. Subsequently, all sequences were aligned and sorted with MAFFT v.7 [34] applying the default settings. If the sequences contained ambiguous characters indicating double infections, the electropherograms of the forward and reverse sequences were carefully rechecked and un-phased using DnaSP v.6.12.3 [35]. This method allows for the identification of lineages in mixed infections if all lineages in the data set are also present in single infections. It requires high-quality electropherograms allowing a clear assignment of double peaks and therefore should be used with caution. In multiple cases, the samples were also screened with the primers CytB_HPL_intF1 (5′-GAGAATTATGGAGTGGATGGTG-3′) and CytB_HPL_intR1 (5′-ATGTTTGCTTGGGAGCTGTAATC-3′) following the protocol by [36], which allows sequencing an 886 bp section of the CytB of all avian haemosporidians. The lineage names of the haplotypes were identified by performing BLAST searches in the MalAvi database (http://​130.​235.​244.​92/​Malavi/​, [10]). New haplotypes identified in the present study were assigned new lineage names and uploaded to the MalAvi database. Haemosporidian sequences isolated from all individual bird samples were uploaded to GenBank.

Of the 310 T. merula individuals, 182 (58.7%) were infected with Plasmodium spp., 15 (4.8%) with Leucocytozoon spp., and 14 (4.5%) with Haemoproteus spp. The most frequent lineages in T. merula were P. matutinum pLINN1 and P. vaughani pSYAT05, both with 97 records (31.3%) each. Other frequent lineages were H. minutus hTURDUS2 with 13 (4.8%), Leucocytozoon sp. lTUMER01 with seven (2.3%), and P. matutinum pAFTRU05 with five (1.6%) infected individuals. Altogether 19 lineages were detected in T. merula, including eight Plasmodium, seven Leucocytozoon, and four Haemoproteus lineages (Table 2). Of these 20 lineages, eight (TUMER12–TUMER18, TUMER20) were first found in the present study. pTUMER12 is closely associated with P. matutinum pLINN1, pTUMER13–pTUMER16 with P. vaughani pSYAT05, hTUMER17 with H. minutus hTURDUS2, lTUMER18 with Leucocytozoon sp. lTFUS14, and lTUMER20 with Leucocytozoon sp. lASOT06. The new lineages each differ in one bp from the already published lineages.

Of the 36 T. philomelos individuals, twelve were infected with Plasmodium spp. (33.3%), six with Leucocytozoon spp. (16.7%), and four with Haemoproteus spp. (11.1%). The most frequent lineages were P. matutinum pLINN1 with eight (22.2%), Haemoproteus sp. hTUPHI01 with four (11.1%), and Leucocytozoon sp. lTUPHI10 and Leucocytozoon sp. lEUSE2 with three (8.3%) positive individuals each. Altogether, eleven lineages were detected in T. philomelos, including each five Plasmodium and Leucocytozoon lineages, and a single Haemoproteus lineage (Table 2). Of these eleven lineages, five (TUPHI08–TUPHI12) were first found in the present study. pTUPHI08 and pTUPHI09 are closely associated with P. matutinum pLINN1, lTUPHI10 and lTUPHI11 with Leucocytozoon sp. lTUPHI04, and lTUPHI12 with Leucocytozoon sp. lEUSE2. The new lineages differ in one bp from the already published lineages, except for lTUPHI10 and lTUPHI11, which differ in three and four bp from lTUPHI04, respectively.

Of the two T. pilaris samples, one was infected with H. minutus TUCHR01.

All sequences were uploaded to NCBI GenBank under the accession numbers MT912098–MT912353, and the new lineages and data were deposited in the MalAvi database. Data on individual birds are also provided in Additional file 1.

Numbers of positive samples are indicated in brackets. The underlined lineage names indicate lineages first detected in the present study.

Based on the ML tree calculated with all Plasmodium sequences available in GenBank (data not shown), three Plasmodium clades were identified, which mainly feature lineages from Turdidae hosts. The data of all lineages clustering in these clades are shown in three separate networks, regardless if they originated from thrushes or other birds. The first network features lineages related to P. vaughani and P. unalis (pTUR1), the second lineages related to P. matutinum and P. lutzi (pTUR2), and the third lineages isolated from American Turdidae (pTUR3). The third clade is closely related to the first one. A fourth clade (pTUR4), featuring only a few lineages isolated from Turdidae birds, is shown for taxonomic reasons, which are discussed later. Detailed information on host species, countries, authors, and publications are provided in Additional file 1. The four DNA haplotype networks feature 56 Plasmodium lineages, which were mainly found in thrushes (Additional file 2). Another 24 lineages, 12 of which are probably specific to thrushes, did not cluster in the networks (Additional file 3). A phylogenetic tree with the lineages featured in the networks and other Plasmodium lineages isolated from thrushes is shown in Additional file 4.

There is only a single clade (hTUR1, Fig. 4) featuring multiple Haemoproteus lineages, which are frequently found in birds of the Turdidae. Besides, also the Haemoproteus majoris clade (hTUR2, Fig. 5) comprises four lineages from North American thrushes, which are not linked to morphospecies yet. Several further lineages attributed or similar to other Haemoproteus species, were found in single specimens of the Turdidae, but are common in other passeriform birds. Detailed information on host species, countries, authors, and publications is provided as Additional file 1. The two DNA haplotype networks feature 17 Haemoproteus lineages, which were mainly found in thrushes (Additional file 5). Another 20 lineages, 11 of which are probably specific to thrushes, did not cluster in the networks (Additional file 6). A phylogenetic tree with the lineages featured in the networks and other Haemoproteus lineages isolated from thrushes is shown in Additional file 7.

Altogether, eight Leucocytozoon clades were identified, which almost exclusively comprise lineages isolated from birds of the Turdidae. The lineages of some clades (lTUR2, lTUR3, lTUR4, lTUR5) were almost exclusively isolated from Turdus spp., whereas those of others (lTUR6, lTUR7, lTUR8) were predominantly isolated from Catharus spp. The network of clade lTUR1 contains lineages isolated from both Turdus spp. and Catharus spp., however, these differ in several bp from each other. Detailed information on host species, countries, authors, and publications are provided in the Additional file 1, and a phylogenetic tree is shown in Additional file 8. The eight DNA haplotype networks feature 94 Leucocytozoon lineages, which were mainly found in thrushes (Additional file 9). Another 25 lineages, 18 of which are probably specific to thrushes, did not cluster in the networks (Additional file 10). A phylogenetic tree with the lineages featured in the networks and other Haemoproteus lineages isolated from thrushes is shown in Additional file 8. The lineages included in the networks have not been linked to a morphospecies yet, except for four parasite lineages studied by [16] in North American birds from Alaska. They identified four parasite lineages in thrushes (lCATGUT02, lTUMIG15, and lTUMIG11), which matched L. majoris morphotypes, and one lineage (lCATMNI01), which matched L. dubreuili morphotypes (Fig. 1 and Additional file 2 in [16]). The three L. majoris-like lineages differ by 4.6 to 10.3% (p-distance) and cluster in three different clades/networks in the present study (lTUR1, lTUR2, and lTUR7). They were also classified as separate species in the multi-gene species delimitation analysis of [16].

This study identified 82 Plasmodium lineages in thrushes, 58 of which clustered into the three clades shown as haplotype networks. Unlike in Leucocytozoon spp., most groups defined in the networks also include lineages isolated from other passeriform birds, which conforms to results of experimental observations showing broad specificity of many avian Plasmodium species [1, 53]. The diversity of Plasmodium lineages is higher in the genus Turdus compared to Catharus, and the networks pTUR1, pTUR2, and pTUR3 do not feature any lineage exclusive to Catharus spp. Only three lineages common in American Turdus spp. (pTUMIG03, pCATUST05, pCATUS06) were isolated from C. ustulatus, and only a single lineage (pCATUST21) was unique to the latter. One reason might be that birds of the latter genus are mainly distributed in the Americas, being less exposed to Old-World parasite lineages. Moreover, the number of Turdus species is considerably higher than that of Catharus with 80 compared to 12 species. The scarcity of Plasmodium spp. might also be explained by the observation that Catharus spp. mostly breed at high-latitude sites in North America where the overall prevalence of Plasmodium is quite low (Spencer Galen, personal communication). However, after this manuscript was submitted, [54] published a study on haemosporidian parasites of Catharus spp. sampled in eastern North America and found seven new lineages in two species of this genus. Two lineages isolated from C. fuscescens (pCATFUS10-11) differ by one and two bp from pCATUST05 (pTUR3, Fig. 2b). Four lineages isolated from C. fuscescens (pCATFUS12-14) and Catharus bicknelli (pCATBIC09) differ in one and three bp from pCATUST06 (pTUR2, Fig. 2a), respectively, and pCATBIC08 isolated from C. bicknelli by ten bp. This new study exemplifies that research in this field is still going on and that screening more bird species and individuals will render additional data.

The 58 Turdidae-specific Plasmodium lineages in the networks may be classified into 17 haplotype groups based on their genetic similarity and geographic and host distribution. Eight groups of similar lineages were found in New World Turdidae compared to nine groups in Old World Turdidae. Only four of the 17 groups of lineages were linked to the morphologically described species P. vaughani (pTUR1 I), P. unalis (pTUR1 IV), P. matutinum (pTUR2 I), and P. lutzi (pTUR2 II). In some cases, the assignation of CytB lineages to morphospecies might be incorrect. To address these issues, data from original species descriptions and further literature were reviewed and are discussed in the following.

Plasmodium (Haemamoeba) matutinum, Wolfson [75] found a Plasmodium parasite in Hylocichla mustelina from Baltimore, Maryland (USA), which she identified as a strain of Plasmodium relictum [formerly Plasmodium praecox]. Experimental infections of canaries showed that the parasite’s biology differed strongly from that of the nominate form of P. relictum, particularly in its strict periodicity and the high degree of synchronism in maturation (segmentation) of erythrocytic meronts [75]. The parasite strain was then isolated from T. migratorius in Kansas, Illinois (USA), and described as Plasmodium relictum var. matutinum by [76], who also confirmed previous findings in further infection experiments with canaries. Huff [76] also tested the susceptibility of Cx. pipiens for Plasmodium relictum var. matutinum and found it to be much lower than for previously tested American P. relictum strains (“Boston” and “Whitmore”). Infection rates for the latter two strains were almost 90% in experimentally fed Cx. pipiens, whereas only 2.5% mosquitos of a mixed culture of Cx. pipiens and Cx. territans were susceptible to Plasmodium relictum var. matutinum [76]. The main vector of the latter is probably Cx. tarsalis, which in experimental studies showed infection rates of up to 84% [77].

Corradetti et al. [78] identified a European strain of P. relictum var. matutinum in the redwing Turdus iliacus in Italy and raised the parasite to species rank. They showed that Cx. pipiens easily could be infected with the European strain, unlike with the North American one [78, 79]. Another difference is the much higher virulence reported for the Italian strain in canaries. Inoculation with the Italian strain was fatal for the canaries and led to death after five to 20 days [78]. However, in canaries experimentally infected with the North American strains, infection built up rapidly for 4 to 5 days but then declined into latency [59]. Nonetheless, susceptibility to mosquito species hardly can be considered as a good taxonomic character because it can be readily changed by selection [1, 59].

Valkiūnas et al. [80] isolated pLINN1 from Luscinia luscinia (Muscicapidae) in Lithuania and linked it to P. matutinum because the morphological characters of the blood stages of pLINN1 corresponded well with those in the parasite’s original description. Experimental inoculations with infected blood showed that domestic canaries are susceptible to pLINN1, but the parasitemia was low and the infection was not lethal, similar to observations on the North American P. matutinum strain [80]. However, only two canaries were exposed, and estimating the virulence is difficult based on the available data. Culex pipiens f. pipiens might be a vector of pLINN1, which was reported in this mosquito, but sporogonic stages were not studied and it remains unclear if the sporogony completes and sporozoites develop in this mosquito species [64, 81]. pLINN1 was found also in 30% of T. merula and 22% of T. philomelos specimens investigated in the present study. The lineage pAFTRU5, which, based on genetic similarity, probably also belongs to the same parasite species, was found in 1.6% of T. merula specimens. Further records in T. merula from Portugal, Western Russia, Armenia, and Morocco [8] indicate that pLINN1 and pAFTRU5 are co-occurring with their avian host species throughout their whole distribution ranges. The data of these studies indicate that P. matutinum pLINN1 and pAFTRU5 are the most common Plasmodium-lineages of T. merula apart from P. vaughani SYAT05. The parasite lineages were also introduced to New Zealand, probably together with their host species [42]. In North America, pLINN1 and pAFTRU05 were recorded only in single specimens of T. migratorius in Michigan [24] and Nebraska [82], respectively. pLINN1 was also detected in Cx. pipiens and Cx. restuans in Ithaca, New York [83].

Based on the present data, T. merula is the main vertebrate host of the lineage pLINN1. This parasite lineage is present but rare in North America, which is the type locality of P. matutinum. Considering the available data about differences in (1) the host and geographic distribution of molecular genetic lineages, (2) the occurrence and susceptibility in mosquitoes, (3) the periodicity and degree of synchronism in maturation (segmentation) of erythrocytic meronts, and (4) the virulence of different isolates in experimentally infected canaries, pLINN1 and related European strains probably do not correspond to the parasite originally described in T. migratorius as P. relictum var. matutinum. pLINN1 probably differs also from the Italian strain studied by [79], because the latter was lethal to canaries. Further experimental studies are needed to test these hypotheses.

Plasmodium (Haemamoeba) giovannolai was described from T. merula in Lazio, Italy, by [84], who reported the parasite only from the type host. Canaries were susceptible and died within a month after experimental infection, and also Cx. pipiens easily could be infected [85]. In biology and morphology, P. giovannolai is remarkably similar to the European strain of P. matutinum depicted in [1], which might be the same strain later linked to pLINN1 and isolated from Luscinia luscinia [80]. However, the pLINN1 isolate from L. luscinia was not lethal in canaries, and is different in this respect from P. giovannolai. Valkiūnas [1] suggests that P. giovannolai might be a subspecies of P. matutinum. Additional support for a close relationship between the two species comes from sequence data of haemosporidian parasites because all Haemamoeba lineages isolated from T. merula (pLINN1, pAFTRU5, pTUMER06, and pTUMER12) exclusively grouped into a single sequence cluster (pTUR2, Fig. 2). Therefore, it might be possible that the original P. giovannolai represents one of the lineages currently considered belonging to P. matutinum. However, despite morphological similarities, the two strains differ notably in one feature. Phanerozoites of P. giovannolai develop almost exclusively in the spleen of canaries but were never seen in the brain, whereas those of P. matutinum are found in numerous organs [1]. A comprehensive study on haemosporidioses of T. merula and T. philomelos confirmed that phanerozoites of P. matutinum pLINN1 develop in multiple organs [86], but tissue stages of pLINN1 were not yet studied in canaries. Additional experimental infections of canaries with pLINN1 could help confirm differences between P. giovannolai and P. matutinum pLINN1 regarding the location of tissue stage formation. It is worth mentioning that the same Plasmodium strain might be markedly different in the ability to produce phanerozoites in the brain of different avian host species [87], so this feature should be used carefully in taxonomic research.

Plasmodium (Haemamoeba) lutzi was described by [88] from Aramides cajaneus (Rallidae, Gruiformes) in São Paulo, Brazil. Gabaldon and Ulloa[89] report the parasite also from birds in Venezuela and suggest that records made by [90] from Columbian birds refer to the same morphospecies. The CytB lineage pTFUS05, isolated from two individuals of T. fuscater in Bogotá, Colombia, was linked to P. lutzi because of morphological similarities [13]. Renjifo et al. [91] linked several additional lineages to P. lutzi, namely pDIGLAF01, pDIGLAF02, pDIGCYA08, and pCATUST05 [original lineage names differ in the publication]. pTUMIG22, isolated from T. migratorius in the USA and T. plumbeus in the Caribbean [22], is also grouped with the latter lineages in clade pTUR2 (Fig. 2). Of these, pCATUST05 is the most common lineage, having been found in several species of American thrushes, birds of other passeriform families, and in the owl species Aegolius acadicus [e.g., 18, 19, 92]. Concerning the identification of pTFUS05 as P. lutzi, there are some taxonomic issues, which deserve attention. The type host of P. lutzi and the host of pTFUS05 belong to separate bird orders, the Gruiformes, and the Passeriformes. Moreover, the type localities São Paulo (Brazil) and Bogotá (Colombia) are more than 4000 km apart from each other and situated in altitudinal zones differing in 2000 meters. [13] also report morphological differences between pTFUS05, the sample studied by [89], and the original P. lutzi. For instance, the original strain of P. lutzi differs from pTFUS05 in various size measurements and particularly in the number of merozoites developed in erythrocytic meronts (16–28 vs. 6–26) [13]. Based on the data summarised in the present study, pTFUS05 probably does not belong to the original strain of P. lutzi. Further studies on Plasmodium parasites of the type host, A. cajaneus, at sites located closer to the type locality would be required to confirm or refute the identity of pTFUS05 as P. lutzi. So far, only a single Plasmodium lineage (pARACAJ01) has been isolated from a specimen of A. cajaneus kept in the São Paulo Zoo [93]. pARACAJ01 differs in about 5% (p-distance) from its closest related Plasmodium lineages and in more than 8% from pTFUS05. Morphological identification of the latter lineage was not possible, because only young stages, mainly meronts, were present in the blood of this specimen (personal communication, Carolina R. F. Chagas).

Plasmodium (Giovannolaia) circumflexum was described by [94] from Turdus pilaris in Germany. This morphospecies was then reported from over 100 bird species belonging to the orders Passeriformes, Anseriformes, Columbiformes, Coraciiformes, Charadriiformes, Falconiformes, Strigiformes, and Galliformes [1]. Valkiūnas et al. [95] linked P. circumflexum to pSW5, a lineage isolated from birds of the Passeriformes, Anseriformes, Gruiformes, and others, but not from Turdidae. They also mention the lineage pTURDUS1 to be closely related to pSW5. pTURDUS1 and pBT7, differing only in one substitution from each other in the CytB barcode section, show similar host distributions as reported for P. circumflexum previously [1] but they differ in 2.3% (p-distance) from pSW5. pTURDUS1 and pBT7 were recorded mainly in passeriform birds, including some Turdidae species (pTURDUS1: T. merula and T. philomelos; pBT7: T. pilaris and T. migratorius), but also in birds of the orders Accipitriformes, Anseriformes, Strigiformes, and Charadriiformes (see MalAvi database). At least pTURDUS1 was also considered a lineage of P. circumflexum by [96] and [62]. This pattern might indicate the presence of cryptic species.

Plasmodium (syn. Proteosoma) tumbayaensis was briefly described by Mazza and Fiora [97] from Planesticus anthracinus, which probably is a synonym of Turdus chiguango, in Tumbaya (Argentina). Blood stages of this parasite are similar to P. vaughani, to which it was synonymised by [98]. Due to the marked genetic diversity of Novyella parasites in Neotropics, validation of this species name is probably possible, but additional research on its biology is needed.

The diversity of Haemoproteus lineages in thrushes is comparably low. In total, 37 Haemoproteus lineages were identified in thrushes, 17 of which clustered in the two networks hTUR1 (Fig. 4) and hTUR2 (Fig. 5). The 17 lineages in the networks can be classified into at least four or five groups belonging to distinct Haemoproteus species. North American thrushes harbor a relatively high diversity of Haemoproteus lineages, with nine lineages in the H. minutus group (Fig. 4), four lineages in the H. majoris group (Fig. 5), and about 20 more rare lineages (mainly from C. ustulatus and South American thrushes) in different clades (Additional file 6). Both the H. minutus and H. majoris clades contain multiple lineages, which were isolated from and complete development with production of gametocytes in non-Turdidae hosts, indicating either relatively recent host switches between Turdidae and other passeriform birds or possible high susceptibility of thrushes to infections, which came from other bird species but abort their development. From this point of view, birds of this group and their Haemoproteus parasites seem to be good model organisms to address abortive development in haemoproteids.

Bennett [99] described Haemoproteus (Parahaemoproteus) fallisi from T. migratorius in Newfoundland, Canada. At the type locality, they found the parasite in 28% of sampled T. migratorius, 22% of Catharus minimus, and 1% of C. ustulatus. Bennett [99] speculated that the same parasite strain might have been recorded in numerous species of Turdidae and Muscicapidae worldwide. Considering that each about 200 specimens of T. migratorius and C. minimus were screened for haemosporidians in recent molecular genetic studies (see MalAvi database), the corresponding CytB lineage should have been detected already. Haemoproteus lineages isolated from T. migratorius and C. minimus cluster in either the H. minutus lineage group (Fig. 4) or the H. majoris group (Fig. 5). In particular, H. majoris shares several morphological characters with H. fallisi: the advanced growing gametocytes both adhere to the erythrocyte nuclei and envelope and slightly displace the nuclei laterally; gametocytes grow around the erythrocyte nucleus, but do not encircle it completely; young gametocytes of H. fallisi are frequently amoeboid, which also applies to H. majoris; gametocytes of H. fallisi contain in average 15 (11–21) and those of H. majoris about 10 pigment granules, which usually are randomly scattered throughout the cytoplasm. However, fully-grown gametocytes of H. fallisi are smaller in length and usually do not reach poles of erythrocytes, a character not characteristic in fully-grown gametocytes of H. majoris. The original H. fallisi strain might, therefore, correspond to one of the North American CytB lineages clustering in the H. majoris group, whose morphological traits should be similar to the original description of H. fallisi and need additional research on the avian type host T. migratorius. To clarify the relationship between H. fallisi and other Haemoproteus species, morphological studies of Haemoproteus parasites of T. migratorius and C. minimus in the type locality (Newfoundland) are suggested. This study also shows that H. minutus is present in T. migratorius and C. minimus but has not been documented in these birds on gametocyte stage level, which needs clarification.

Haemoproteus (Parahaemoproteus) majoris was described by [100] from Parus major (Paridae) in Metz (France). The parasite was reported from more than 80 passeriform species in the Holarctic, Ethiopian, and Oriental regions [1]. Most host species belong to the families Paridae, Sylviidae, Muscicapidae, and Fringillidae, but Turdidae were not mentioned as vertebrate hosts of H. majoris. However, four lineages (hPHYBOR04, hPOEATR01, hTUMIG08, and hTUMIG21), isolated from the North American thrushes T. migratorius, C. minimus, and C. guttatus, differ only in few bp from lineages currently attributed to H. majoris. All other lineages in the H. majoris group (Fig. 5) were isolated from non-Turdidae passeriform birds in Western Europe, Eastern Europe, and Africa. There is only a single record of hPARUS1 from T. merula in Armenia [8], which might be a case of an abortive infection. Nilsson et al. [45] linked several lineages to H. majoris (hPARUS1, hPHSIB1, hWW2, hCCF5, and hCWT4) and discussed that these might belong to a complex of cryptic species. The North American lineages differ in one to a few bp from those linked to H. majoris and should be considered as part of the same species group. Detailed morphological studies of these lineages are required to support the latter assumption.

Cleland and Johnston [101] described Haemoproteus (Parahaemoproteus) geocichlae from Zoothera lunulata in New South Wales, Australia. The parasite has not been reported from other host species and sampling sites since. Beadell et al. [7] screened samples of 105 bird species from Australia and Papua New Guinea and detected hZOOLUN01 in one of two Z. lunulata specimens. hZOOLUN01 differs from H. minutus hCOLL2 in one substitution and is the only Haemoproteus lineage isolated from Z. lunulata so far. Beadell et al. [7] mention that they prepared blood films from this bird but did not publish pictures of the parasites. To clarify the identity of this parasite lineage, morphological studies of gametocytes in Z. lunulata in Australia and the designation of type material are required.

The Haemoproteus minutus group includes several parasite species, which are common in thrushes: H. minutus, H. pallidus, and H. homominutus. Interestingly, the CytB lineages of these species differ only in a few bp from each other but show different host preferences and distinct morphological features both at gametocyte and sporogony stages. Most lineages of these species have Holarctic distributions, but some lineages of this species group were found also in Australia (GERPAL02, PARPUN01, SERCIT02, and ZOOLUN01), sub-Saharan Africa (AFR076, AFR081, BERMAD02, ILCLE01, TUROLI05, XANZOS01, and XANZOS02), and South America (RAMCAR06) (Fig. 4).

Haemoproteus (Parahaemoproteus) minutus was described by [102] from T. merula sampled at the Curonian Spit in Kaliningrad District, Russia. Palinauskas et al. [44] linked hTURDUS2, hTUCHR01, and hTUPHI01 to H. minutus because of their genetic similarity and common morphological traits. Differences in host composition indicate that hTUPHI01 might belong to a separate, cryptic species. Whereas hTURDUS2 was mostly isolated from T. merula, hTUPHI01 was almost exclusively isolated from T. philomelos (Fig. 4). Although the distribution ranges of T. merula and T. philomelos vastly overlap in Europe and Western Asia, the two bird species are not closely related and therefore might be prone to infection with different parasites. Moreover, the lineages hTUPHI01 and hTUCHR01/hTURDUS2 are not directly connected in the network, but via the intermediate haplotype hCOLL2, which belongs to a morphologically distinct parasite species, H. pallidus. hTUCHR01 also differs from hTURDUS2 in geographic and host distribution, being more common in Russia and Asia.

Haemoproteus (Parahaemoproteus) homominutus was recently described from T. viscivorus in Lithuania by [46], who reported the parasite also from T. merula. The Lineage hCUKI1 was so far only isolated from the type host T. viscivorus in Lithuania and Morocco, and Culicoides circumscriptus in Turkey. The records from C. circumscriptus were published in GenBank (KX831074 und MF095643) only, and the vector competence for H. homominutus has not been tested yet. The lineage hCUKI1 has not been isolated from other Turdidae species and might preferably infect T. viscivorus, although it was also seen in T. merula on gametocyte stage [46].

Valkiunas and Iezhova [103] described Haemoproteus (Parahaemoproteus) pallidus from Ficedula hypoleuca (Muscicapidae) in Kaliningrad District, Russia. According to [1], the distribution of H. pallidus is insufficiently investigated but includes the Palearctic and the Ethiopian zoogeographical regions. [104] linked the CytB lineages hCOLL2 and hPFC1 to H. pallidus. The main hosts of hCOLL2 and hPFC1 are the two Muscicapidae species Ficedula albicollis and F. hypoleuca with more than 300 individual records of each lineage, whereby hPFC1 is more common in F. hypoleuca and hCOLL2 in F. albicollis [105‐107] (sequences of the latter studies were not published on GenBank and therefore not included in the DNA haplotype network in Fig. 4). Both Muscicapidae species are long-distance migrants with breeding habitats in Europe and Western Asia and wintering habitats in Sub-Saharan Africa. In Cameroon and Gabon, hCOLL2 was also found in the white-tailed ant thrush Neocossyphus poensis [108], but there is only a single record from T. merula in Western Russia [8]. However, in Northern and Central America hCOLL2 was almost exclusively isolated from thrushes, namely T. migratorius, T. assimilis, C. minimus, and C. ustulatus. Concluding, H. pallidus is not only distributed in the Palearctic and the Ethiopian regions, but also the Nearctic. Interestingly, there are still no reports about the presence of gametocytes of H. pallidus in American thrushes. Nevertheless, microscopical blood slide analyses are required to confirm that North American thrushes are competent hosts for hCOLL2 and related lineages and that the parasites develop gametocytes in thrushes. The lineages hCATUST22, hTURASS04, hTUMIG07, hTUROLI05, and hZOOLUN01, which are linked to hCOLL2 via one or two substitutions, were also isolated from thrushes, but the morphological identification of these parasites has not been developed yet.

Regarding the diversity of CytB lineages, Leucocytozoon represents by far the most diverse genus of haemosporidians in thrushes with 119 lineages, 94 of which clustered into the eight clades shown as haplotype networks. Most of these lineages were exclusively isolated from birds of the family Turdidae and therefore seem to be specifically adapted to this host group. Moreover, unlike Plasmodium and Haemoproteus parasites, the Leucocytozoon lineages are almost exclusively restricted to either species of the genus Turdus or Catharus, but they are usually not shared between them. The clades lTUR2 (Fig. 6b), lTUR3 (Fig. 7a), and lTUR4 (Fig. 7b) contain almost exclusively lineages isolated from Turdus spp., and lTUR5 (Fig. 7c), lTUR6 (Fig. 8a), and lTUR7 (Fig. 8b) from Catharus spp. Only clade lTUR1 (Fig. 6a) contains Leucocytozoon lineages isolated from species of both bird genera, however, in distant subclades.

The 94 Turdidae-specific Leucocytozoon lineages in the networks may be classified into at least 38 groups, which potentially belong to distinct species, but only a few have been linked to morphospecies yet. [16] identified four parasite lineages in thrushes (lCATGUT02, lTUMIG15, and lTUMIG11), which matched L. majoris morphotypes, and one lineage (lCATMNI01), which matched L. dubreuili morphotypes (Fig. 1 and Additional file 2 in [16]). However, the three L. majoris-like lineages and several additional ones isolated from other bird families (hACAFL01, hBT1, hCB1, hJUNHYE04, and hPHYBOR1) are genetically distinct and were mostly classified as separate species in the multi-gene species delimitation analysis of [16]. These lineages might represent cryptic species sharing similar morphological traits.

The majority of these Leucocytozoon lineage groups (25) was isolated exclusively from thrushes in the Americas, compared to eleven from Old World Turdus species. Only two groups (lTUR1 II and lTUR3 I) contain lineages from both New-World and Old-World Turdus species. The diversity was particularly high in North American thrushes. In South America, Leucocytozoon spp. were only isolated from thrushes native to the Andes Mountains (T. fuscater, T. falcklandii, T. serranus, T. nigriceps, T. chiguanco, and C. fuscater). This is in accord with literature reporting Leucocytozoon to be generally rare in the Neotropical region, which might be related to the absence of suitable vectors in lowland areas of South America [1]. Simuliidae black flies are present in South American tropical lowland areas [e.g., 109, 110], but the local species might not be susceptible to Leucocytozoon spp. [111] summarised records of Leucocytozoon spp. in South America and also present a case of Leucocytozoon infecting a host population in Amazonia. They suggest that high average temperatures may be constraining the diversity of Leucocytozoon in lowland tropical South America, because parasites of this genus generally require lower temperatures for the development in their blackfly vectors [111]. A general problem in the classification and identification of Leucocytozoon species is the low number of distinctive morphological characters in their blood stages. Gametocytes of Leucocytozoon parasites do not produce pigment granules, whose number, morphology, and location are important morphological traits for Plasmodium and Haemoproteus species. Unlike Plasmodium parasites, Leucocytozoon species do not multiply in the blood and do not produce blood-cell meronts, which provide additional morphological features for malaria parasite classification. Consequently, the main criteria for delimiting Leucocytozoon species have been the shape and location of infected host cell nuclei and the morphology of host cell-parasite complexes (roundish or elongate, with different shapes of the cytoplasmic processors in the latter). Additionally, the occurrence in different host groups was also taken into consideration during Leucocytozoon species description, assuming that host switches between different bird families and orders represent unlikely or rare events. Therefore, morphologically similar parasites isolated from birds of separate orders or families were often considered distinct species [1, 53]. These morphological characters certainly have taxonomical value, and many readily distinguishable Leucocytozoon species exist and were described, but the marked overlap of main diagnostic features in different Leucocytozoon species is also evident. This study supports the existence of cryptic speciation in the genus Leucocytozoon. This is not surprising due to (1) the poor morphological differentiation of leucocytozoid blood stages being readily accessible for research, and (2) because of the lack of data about other stages of development for the majority of described species. It is probable that many more Leucocytozoon species exist, but they hardly can be distinguished solely based on the blood stages of the parasites. Examination of vectors and parasite sporogonic and tissue stages might be needed to morphologically characterise these species. The difficulties in morphological identification may explain, why only a few Leucocytozoon species were described in the last decades, and only little progress was made regarding linking CytB lineages to morphospecies.

To go ahead with the molecular characterization of Leucocytozoon parasites, the identification of CytB lineages of the most common species parasitizing passerines (Leucocytozoon fringillinarum, L. majoris, L. dubreuili, and others) are needed. These parasite species names have taxonomic priority due to their early dates of description [112], but their lineages remain unidentified from the type hosts and type localities. The lack of this information precludes further taxonomic work on leucocytozoids in passeriform birds. The main obstacle of linking parasite morphotypes with their lineages is that co-infections are common and even predominate, being visible in blood films but not always detectable or distinguishable using standard PCR-protocols [49]. In the case of Leucocytozoon parasites this problem seems to be even more severe than in the case of Plasmodium and Haemoproteus infections [48], but still remains insufficiently addressed. After this manuscript was submitted, [54] published a study for which they screened 460 individuals of C. minimus, C. fuscescens, and Catharus bicknelli for haemosporidian parasites. They identified 16 new Leucocytozoon lineages (lCATBIC01–06, lCATBIC10–13, lCATFUS01–02, lCATFUS04–09, and lCATMIN11) in these three bird species, which were not included in the present study. All of these lineages would cluster in the Leucocytozoon networks presented here, differing from already published lineages by one to six bp. This study exemplifies that this area of research is still developing and that sampling of larger numbers of individuals and additional host species will refine our understanding on haemosporidians of thrushes.

Leucocytozoon dubreuili was briefly described by [113] from Turdus sp. in Tonkin, Vietnam. The French name of the type host is “grive”, which translates to “thrush”. More than 15 Turdidae species are known from Vietnam, half of which include “grive” in their French name (Birds of Vietnam; http://​sibagu.​com/​vietnam/​turdidae.​html), wherefore the type host cannot be deduced from the original description. Later, the parasite was reported from over 60 species of passeriform birds from the Holarctic, Ethiopian, and Oriental regions, but there are only a few records from the Neotropical and Australian regions. Moreover, sporogonic development of L. dubreuili was reported in several Simuliidae species belonging to the three different genera Cnephia, Prosimulium, and Simulium [1]. As long as the vertebrate type host is unknown, linking the species to a CytB lineage would be afflicted with severe uncertainties. Although morphological studies of haemosporidians of South-East Asian and Southern Asian birds are available (e.g., [114]), these regions still are black spots in regard to diversity of CytB lineages. These regions harbor a large variety of Turdidae species from different genera, but almost no sequence data of their haemosporidian parasites are available. Moreover, it seems unlikely that records of L. dubreuili from non-Turdidae passeriform birds refer to the same parasite species because Leucocytozoon lineages of Turdidae usually seem to be restricted to members of the same family or species (data of the present study). Two other Leucocytozoon species were also described from Turdidae hosts but were later synonymised with L. dubreuili based on the morphological identity of their blood stages and host cells by [115]. Leucocytozoon giovannolai was described from T. iliacus in Northern Africa by [116], and Leucocytozoon mirandae from T. merula in Portugal by [117]. The latter two taxa were isolated from different host species and geographic regions, and therefore might be distinct from the parasite strain originally described as L. dubreuili. These species names might be validated if blood samples from type hosts and their CytB sequences are available.

Leucocytozoon maccluri was described by [118] from Zoothera marginata in Thailand. The parasite was found in two specimens of the type host but was never reported from other host species. The morphology of the gametocytes and their host cells is unique and differs strongly from that of other Leucocytozoon species. Unfortunately, haemosporidian sequences of Z. marginata were not published so far.

The MalAvi database was established by [10]. It first promoted the use of a 478 bp section of the mt CytB as a standard DNA barcode sequence and introduced a unified naming scheme for avian haemosporidian lineages. Moreover, it summarises information on geographic and host distribution of CytB lineages and provides references to the original publications and parasite descriptions, including details on their morphological characters [10]. Importantly, the sequences of new lineages deposited in MalAvi database are checked for their quality and are minimised in regard to possible misidentifications, which remain an insufficiently controlled problem in GenBank [119]. However, a sensitive problem for future research concerning MalAvi database is that authors of many publications made the data only available in the MalAvi database, but did not submit the sequences to GenBank or other government-financed sequence repositories. The only CytB sequences contained in MalAvi database are those in the “Grand Lineage Summary” table, which features a single sequence for every unique lineage, but the “Host and Sites” table, which summarises information on host species, localities, and references, contains lineage names only. Unless the sequences were not submitted in parallel to GenBank or other platforms, information on sequence length and quality is lost. Moreover, these records will not show up in GenBank searches, which is disadvantageous for future research, and which was also a problem for the data analysis of the present study. To ensure that records stay permanent, it is recommended to submit sequences of all individual samples (including information on host species, geographic origin, age, etc.) to GenBank. Alternatively, it is possible to submit one sequence per lineage detected to GenBank and provide data for individual host specimens either in the publication or as supporting information (e.g., [8]). This would allow retrieving the data with GenBank searches and at the same time would facilitate incorporation into the MalAvi database. The GenBank submission process currently also involves sequence checks, which detect interruptions of the coding frame and other errors, which helped to improve the quality of sequences in more recent GenBank entries. Concluding, the authors of the present study strongly encourage researchers to deposit their highly valuable data to GenBank in all cases. This will open opportunities to more precise research of the sort presented in this study as well as enrich the data available in MalAvi.