Molecular phylogenetic approaches continue to uncover unanticipated diversity in avian haemosporidian species of Plasmodium, Parahaemoproteus, and Leucocytozoon (Ricklefs and Fallon, 2002; Bensch et al., 2004; Sehgal et al., 2006; Martinsen et al., 2008). The diversity of these parasites seems to match that of the avian hosts, and the raptorial Falconiformes (hawks and falcons) and Strigiformes (owls) appear to be no exception. Hawks and owls are prone to infection by Leucocytozoon (Leucocytozoon toddi and Leucocytozoon danilewski, respectively), but also by species of Plasmodium and Parahaemoproteus. The most recent morphologic–taxonomic assessment of avian haemosporidian diversity (Valkiunas, 2005) lists 6 Parahaemoproteus species unique to Falconiformes and 2 unique to Strigiformes; within the more species-rich Plasmodium, only 3 are unique to Strigiformes, while 3 additional species infect both Strigiformes and Falconiformes as well as many other host taxa. As in the case of most other host taxa (Bensch et al., 2000; Ricklefs et al., 2004), molecular characterizations of the haemosporidian parasites of these groups of raptors reveal many lineages distributed widely within the haemosporidian phylogeny which may or may not be unique to raptorial birds.

Four recent molecular studies highlight patterns of host distribution and phylogenetic relationships among raptor parasites. The morphologically defined species L. toddi is unique to Falconiformes and is geographically widespread, yet Sehgal et al. (2006) showed that this “species” has distinctive mitochondrial DNA sequences in Accipiter and Buteo spp. in both the New World and the Old World. Two parasite clades within the morphologically defined L. toddi were 10.9% divergent in mitochondrial cytochrome b (cyt b) and, because this divergence is coupled with host specificity within and among clades, the authors suggest that the cyt b lineages represent cryptic species.

Ortego et al. (2007) focused on Plasmodium and Parahaemoproteus parasite species from 1 host species, Falco naumanni (lesser kestrel), and confirmed 3 morphologically described species from both parasite genera, as well as other undescribed lineages identified by unique cyt b sequences, particularly of Parahaemoproteus. None of the recovered sequence-based lineages was of parasites unique to Falconiformes. In another Old World study having a larger host-taxonomic scope, Krone et al. (2008) identified 1 Parahaemoproteus lineage to species, P. noctuae, a morphologically defined parasite unique to Strigiformes. Other raptor haemosporidian lineages were interspersed phylogenetically among the parasites of passeriform (perching; passerines) birds, while 1 unidentified lineage on a very long branch was unaffiliated with either Parahaemoproteus or Plasmodium.

Lastly, Ishak et al. (2008) found many lineages of Leucocytozoon, Plasmodium, and Parahaemoproteus among a variety of strigiform (owl) host species. Overall, these studies highlight an unsurprising lack of phylogenetic cohesion among raptor parasites. Considering that there is no a priori reason to suspect that parasite lineages would be unique to these highly divergent orders of avian hosts (Hackett et al., 2008), a more wide-ranging assessment of raptor parasite diversity is in order.

In the present study, we include Parahaemoproteus and Plasmodium cyt b “lineages” from 43 falconiform and strigiform host samples and evaluate these with previously published data on bird, lizard, and mammal parasites and with data generated in our laboratory for avian parasites from raptors, passeriform, columbiform (doves and pigeons), and anseriform (ducks and geese) birds. We augment these cyt b data for many of our samples with sequence data from the mitochondrial cytochrome oxidase 1 (cox I), apicoplast caseinolytic protease C (ClPC), and nuclear diacylglycerol-O-acyltransferase (trf) genes. Our main objectives were to characterize the diversity and relationships among raptor parasites and to identify clades unique to raptors as a whole, or to groups or individual species of raptors. We did not attempt to relate lineages defined by DNA sequences to described species of parasites. A significant finding of this study is a new clade of haemosporidian parasites that is differentiated from both Plasmodium and Parahaemoproteus.

MATERIALS AND METHODS

Laboratory procedures

DNA extraction and polymerase chain reaction (PCR)-based screening for parasites were conducted as in Fallon et al. (2003). Of those screened, 29 individuals were positive for malaria parasites (Table I). PCR for partial cyt b sequence data was performed on positive samples using primers 543F and 926R (∼300 base pair [bp]) as in Fallon et al. (2004); these products were purified and then sequenced with the same primers on an ABI 3100 (Applied Biosystems, Foster City, California) using standard protocols. For a subset of these samples (n  =  14), full-length cyt b sequences (∼1,100 bp) were obtained using primers and protocols from Perkins et al. (2007; see  Fig. S1 (10.1645_GE-1982.1.s1.doc), available online). Additional sequences from the cox I (∼1,100 bp), ClPC (∼600 bp), and trf (∼300 bp) genes were generated with primers from Perkins et al. (2007; cox I and ClPC) and Beadell et al. (2009; trf); for 12 samples (3 raptors and 9 passerines), we amplified partial cyt b using primers HaemF and Haem R1 (Hellgren et al., 2004).

Table I

Raptor cytochrome b sequences included in phylogenetic analyses. Sample name: the parasite lineage name (Fig. 1); STRI: Strigiformes host, FALC: Falconiformes host. parasite genera identified: parasite lineages clearly associated with either Plasmodium or Parahaemoproteus were assigned as appropriate from Figure 1, whereas Unknown refers to the divergent parasite clade (Clade 7, Fig. 1). Samples with multiple infections are noted. Host species, geographic locality of sample collection, clade (from Fig. 1), and accession number (2, if applicable) are listed for each sample. Accession numbers EU627829 through EU627845 are from Ishak et al. (2008).

Mitochondrial sequences were edited and aligned unambiguously using BioEdit (Hall, 1999). All sequences translated into appropriate codons, as compared to the Plasmodium falciparum cyt b gene (Gardner et al., 2002), and no stop codons were detected. We combined our sequences with a larger dataset, including additional avian parasite samples from our own laboratory and a diverse sample of avian, mammal, and lizard parasites from GenBank (Appendix 1, available online); including these diverse parasites is necessary given the non-monophyly of avian malaria parasites (Martinsen et al., 2008). With 3 Leucocytozoon outgroup species, the cyt b dataset comprised 131 samples; additional Leucocytozoon sequences did not alter the results in preliminary analyses and, therefore, only 3 samples were used to reduce computation time. Limited data were included from previous studies due to non-overlap of sequences from those studies with sequences obtained from 1 primer set (543–926) in this study which highlighted an important, novel clade (see Clade 7 below, and cyt b data map in Perkins et al. 2007).

RESULTS

For 3 samples (Table I), the 543–926 and full-length cyt b primers amplified different sequences. In BLAST searches (Altschul et al., 1990), the 543–926 sequences obtained for 1 clade were no more than 82% similar to known haemosporidian cyt b sequences from Leucocytozoon, Parahaemoproteus, and Plasmodium. These sequences are hereafter referred to as raptor cyt b Clade 7 (Fig. 1). Other 543–926 sequences, as well as the full-length cyt b sequences, were 99% similar to either Parahaemoproteus or Plasmodium sequences (not shown). Inclusion of data from both 543–926 and full-length cyt b primer sets indicates that Clade 7 is not closely related to other haemosporidian parasite clades, i.e., it is no more closely related to Leucocytozoon lineages than it is to Parahaemoproteus or Plasmodium lineages (Fig. 1).

Figure 1

Maximum likelihood cytochrome b phylogeny of haemosporidian parasite lineages. Maximum likelihood bootstrap values above 0.75 are listed above nodes and Bayesian support values above 0.75 are listed below nodes.

Ignoring the placement of Clade 7 for now, the remaining raptor parasite lineages fall into either Parahaemoproteus or Plasmodium (Fig. 1; Table I); most of the sequences from this study, and from previously published raptor parasite lineages (Ishak et al., 2008), are closely related within raptor-specific clades. Within Parahaemoproteus, all but 2 raptor parasites fall into 2 raptor-specific, monophyletic clades. One of these, Clade 13, has very high support and is composed of parasites recovered from Falco sparverius (American kestrel) and Tyto alba (barn owl), while the other, Clade 9, almost exclusively includes owl parasites, with the exception of 1 Buteo jamaicensis (red-tailed hawk) parasite. The latter clade is closely related to Clade 8, with 7 raptor sequences, but also contains a duck parasite, Parahaemoproteus enucleator, and a parasite associated with Coraciiformes (kingfishers). The remaining 2 raptor Parahaemoproteus parasites, both from Buteo jamaicensis, are members of a clade including passerine parasites (Clade 12; Fig. 1). Within Plasmodium, raptor parasites are not monophyletic, but instead are distributed across several clades (Clades 3, 4, and 6; Fig. 1).

Returning to Clade 7, all members are parasites of Buteo species in both North America and Europe, with the exception of 1 sample from a Circus aeroginosus (marsh harrier) of Eurasia. Full-length cyt b sequences from 3 of these samples differed from the Clade 7 sequences obtained with the 543–926 primers; the full-length sequences align closely with lineages from either Parahaemoproteus (Clade 12, 2 cases) or Plasmodium (Clade 4, 1 case; Table I;  Appendix S1 (10.1645_GE-1982.1.s1.doc)). There is nothing unusual about these primer sequences when comparing them to all raptor parasite sequences. Primer 543F exhibits 2 polymorphic sites; in both cases, the polymorphism is restricted to Parahaemoproteus, and both Plasmodium and Clade 7 lineages are identical in this region. Primer 926R contains no polymorphic sites across sequences that overlap. In addition, sequences from 3 raptor samples generated with primers HaemF and HaemR1 also closely aligned with Plasmodium (Clade 3, 1 case;  Appendix S1 (10.1645_GE-1982.1.s1.doc)) and Parahaemoproteus (Clades 2 and 13, 2 cases;  Appendix S1 (10.1645_GE-1982.1.s1.doc)). All passerine sequences generated with full-length cyt b primers (Perkins et al., 2007) and HaemF–HaemR1 (Hellgren et al., 2004) were identical (data not shown).

Contamination of these samples, whether at extraction or during PCR, seems unlikely; the samples in Clade 7 were extracted at different times and, when they were extracted, other samples clearly held either Plasmodium or Parahaemoproteus parasites. The unusual Clade 7 sequences were repeatable with the 543–926 primers, but other primer sets (from our laboratory and from the literature) were unable to amplify additional, i.e., longer, cyt b sequence data from this unusual “lineage.” Moreover, cox I, ClPC, and trf data all suggest that this clade is divergent from other malaria parasite lineages; Clade 7 samples did not group with either Plasmodium or Parahaemoproteus in phylogenetic analyses based on data from these 3 other markers (see  Fig. S2a, b, c, available online (10.1645_GE-1982.1.s1.doc)).

DISCUSSION

Our broad sample of raptor haemosporidian parasites has revealed 2 raptor-specific clades (9 and 13) and 1 raptor-dominated clade (8) within Parahaemoproteus, several parasite lineages scattered within the Plasmodium phylogeny (Clades 3, 4, and 6), plus an additional, previously unrecognized clade that forms a polytomy with Plasmodium and Parahaemoproteus (Clade 7). Two clades exclusively contain raptor parasites. Even the Parahaemoproteus clades of raptor parasites (9 and 13) exhibit host specificity to some degree, i.e., Clade 9 includes exclusively owl parasites (with 1 exception) while Clade 13 includes only American kestrel and barn owl parasites. In addition, 2 Parahaemoproteus parasites recovered from red-tailed hawk were from Clade 12, which contains a wide variety of parasites of passerines. Within Plasmodium, parasites of raptors do not form exclusive clades, but are distributed in Clades 3, 4, and 6 with the parasites of a wide variety of other bird groups (Fig. 1). Thus, several Parahaemoproteus lineages might be raptor specialists. The exclusively raptor Clade 7 is not clearly associated with either Plasmodium or with Parahaemoproteus, but rather forms a polytomy with these genera.

The ML phylogenetic analysis does not clearly associate Clade 7 with either Plasmodium or Parahaemoproteus. Krone et al. (2008) found a similarly unplaced parasite lineage that belongs with our Clade 7 (see below). Clade 7 is notable for several reasons. First, in our study, sequences for this clade were generated only by a single set of cyt b primers (543–926). These are the only cases from more than 100 samples in which 2 primer sets for cyt b have not produced identical sequence (this study; D. Outlaw and R. Ricklefs, unpubl. obs.). Second, the branch leading to this clade is longer than any other branch in the phylogeny, owing to a higher proportion of amino acid changes (Fig. 1; 13 synonymous and 17 nonsynonymous nucleotide substitutions; see also  Appendix S2 (10.1645_GE-1982.1.s1.doc) for amino acid differences). Third, with the exception of 1 Circus aeroginosus sample and 1 Buteo buteo (common buzzard) sample from Spain, the remaining parasite sequences were obtained from North American Buteo samples. It is common for raptor parasite lineages to be found in both the New and Old Worlds (Sehgal et al., 2006), but the host specificity in North America is striking; further sampling of Old World raptors may reveal additional raptor hosts that harbor this parasite lineage. In the case of the divergent B. buteo parasite from Germany reported by Krone et al. (2008), within the small region of overlap, Clade 7 Buteo sequence data and the Krone et al. (2008) sample are identical in many cases and no more than 2% divergent, whereas across all lineages, divergence values range from 3 to 28%. We suspect that additional data will confirm that this parasite belongs to Clade 7, as does a parasite recovered from B. buteo in Spain (FALC16; Table I, Fig. 1).

Amplification of different sequences with different primer sets has several possible explanations including contamination, the amplification of divergent copies of cyt b within the parasite's concatenated mitochondrial genome, the presence of nuclear copies of cyt b, and multiple infections (which we found in 1 case, i.e., STRI15; Table I). Because these parasites were consistently amplified in many independent trials, we believe we can discount contamination. Morphological identification of these parasites might provide some clarification, but blood smears were not available from these samples.

With regard to amplification of different cyt b copies, it is pertinent that the mitochondrial genome of haemosporidian parasites, which contains only 3 genes, i.e., cyt b, cox I, and cox III as well as several remnant tRNAs and rRNA fragments (Vaidya and Arasu, 1987; Vaidya and Mather, 2005), is present in multiple, concatenated copies. Accordingly, different primer sets might amplify cyt b copies from different copies of the mitochondrial genome that have undergone divergent evolution. However, this is unlikely simply because, while Clade 7 lineages are monophyletic, i.e., have shared ancestry, sequences generated for the same samples with other primer sets are not—they fall into either Plasmodium or Parahaemoproteus (Table I; Fig. 1). If Clade 7 lineages represent alternative cyt b copies, then we would expect these lineages to have been derived from, i.e., evolved from, the functional copies of the respective lineages to which the parasites seem to belong (from other primer sets), or to have been equally divergent from each other as they are from Parahaemoproteus and Plasmodium sequences.

Alternatively, the divergent cyt b sequence may be a non-transcribed nuclear insert, or numt (Richly and Leister, 2004). However, this partial cyt b sequence contains no stop codons, which suggests that the integrity of the protein is being maintained. Moreover, the “alternative cyt b copy” argument articulated above also makes numts unlikely as well.

Lacking plausible alternatives, multiple infections seem the most likely explanation for the presence of widely divergent, alternative cyt b sequences in several of the samples. However, none of our sequences revealed double peaks; thus, single primer sets did not amplify sequences from more than 1 lineage within the same host. Although Clade 7 differs substantially from the other lineages, other primer sets amplify cyt b sequences across the entire range of Parahaemoproteus and Plasmodium. Multiple infections are commonly detected across avian malaria parasite groups (e.g., Perez-Tris and Bensch 2005; D. Outlaw and R. Ricklefs, unpubl. obs.; D. Santiago-Alarcon et al., unpubl. obs.). Additional data from other markers obtained from Clade 7 cyt b lineages are likewise highly divergent from other parasite data, but this does not help to clarify the affiliation of this parasite group (see  Fig. S2a, b, c (10.1645_GE-1982.1.s1.doc)).

Even with the current sampling of Leucocytozoon, Plasmodium, Haemoproteus, and Parahaemoproteus, we cannot place Clade 7 within any of these other parasite genera. It is clear that Clade 7 is associated with Plasmodium and Parahaemoproteus from data for cyt b (Fig. 1), cox I, and ClPC (see  Fig. S2a, b (10.1645_GE-1982.1.s1.doc)). Lacking outgroup taxa for trf data (see  Fig. S2c (10.1645_GE-1982.1.s1.doc)), we cannot determine the clade's affiliation with this marker. It is surprising that all 3 other markers, i.e., cox I, ClPC, and trf, seem to amplify the same novel lineage as the 543–926 cyt b primer set. More cox I, ClPc, and trf data from raptor parasites, as well as a broad screening of raptor parasites using alternative primer sets of these markers, may shed some light on the phylogenetic position of this clade.

Raptor parasites are extremely diverse, belong to raptor host-specific clades in Parahaemoproteus but not in Plasmodium, and include a highly divergent, novel parasite lineage (Clade 7). An intriguing result was the dual amplification of cyt b sequences belonging to this divergent lineage and to conventional lineages within the same host sample. The highly divergent cyt b sequences are supported by equally divergent cox I lineages in the mitochondrial genome that place Clade 7 outside of Parahaemoproteus and Plasmodium and by an apicoplast marker and a nuclear marker that are consistent with the existence of a highly divergent parasite lineage. Although we offer several possible explanations for the dual amplification of the cyt b gene, we argue that multiple infection is the most likely explanation. Further investigation of raptor parasites and complete sequencing of mitochondrial genomes may offer insight into the patterns we present here and help to elucidate the taxonomic identity of lineages within Clade 7.