Introduction
Emerging infectious diseases are caused by parasites and pathogens that appear in new areas of the world, or in new host species, and can represent serious threats to biodiversity and human health (Daszak et al., Reference Daszak, Cunningham and Hyatt2000). For example, the fungal pathogen of bats Pseudogymnoascus destructans (white-nose syndrome), recently appeared in eastern North America, possibly via human introduction from Europe, and has devastated bat populations there (Frick et al., Reference Frick, Pollock, Hicks, Langwing, Reynolds, Turner, Butchkoski and Kunz2010, Reference Frick, Puechmaille, Hoyt, Nickel, Langwig, Foster, Barlow, Bartonička, Feller, Haarsma, Herzog, Horáček, van der Kooij, Mulkens, Petrov, Reynolds, Rodrigues, Stihler, Turner and Kilpatrick2015). Many similar negative effects of pathogens on evolutionarily naïve hosts have been documented (e.g. West Nile Virus impact on birds after introduction to North America (LaDeau et al., Reference LaDeau, Kilpatrick and Marra2007); the emergence of the chytrid fungus Batrachochytrium dendrobatidis as a major amphibian pathogen (James et al., Reference James, Toledo, Rödder, Silva Leite, Belasen, Betancourt-Román, Jenkinson, Soto-Azat, Lambertini, Longo, Ruggeri, Collins, Burrowes, Lips, Zamudio and Longcore2015), etc.). Despite the importance of determining where emerging infectious diseases originate, the biogeography of few major taxonomic groups of parasites and pathogens has been documented (Gaunt et al., Reference Gaunt, Sall, de Lamballerie, Falconar, Dzhivanian and Gould2001; Morand and Krasnov, Reference Morand and Krasnov2010; Summerell et al., Reference Summerell, Laurence, Liew and Leslie2010; Pagenkopp Lohan et al., Reference Pagenkopp Lohan, Fleischer, Carney, Holzer and Ruiz2016). Biogeographic analyses of whole clades may help reveal the frequency and geography of natural parasite and pathogen dispersal and provide information in support of modelling and prediction of emerging infectious diseases. Such analyses might also indicate the relative contribution of divergence in allopatry, as opposed to different hosts, in generating pathogen diversity.
Avian haemosporidian parasites (order Haemosporida) of the genera Plasmodium, Haemoproteus and Leucocytozoon are widespread in birds on all continents except Antarctica (Valkiūnas, Reference Valkiūnas2005). These parasites are transmitted by dipteran vectors [Plasmodium by mosquitos (Culicidae), Haemoproteus by biting midges (Ceratopogonidae) and hippoboscid flies (Hippoboscidae), and Leucocytozoon by black flies (Simuliidae) (Valkiūnas, Reference Valkiūnas2005)], in which the parasites develop and undergo sexual reproduction. Individual species of parasite often infect a broad range of avian host species (Hellgren et al., Reference Hellgren, Perez-Tris and Bensch2009; Ricklefs et al., Reference Ricklefs, Outlaw, Svensson-Coelho, Medeiros, Ellis and Latta2014) and many parasite lineages have switched frequently to new avian hosts over the course of their evolutionary history (Ricklefs et al., Reference Ricklefs, Fallon and Bermingham2004, Reference Ricklefs, Outlaw, Svensson-Coelho, Medeiros, Ellis and Latta2014; Alcala et al., Reference Alcala, Jenkins, Christe and Vuilleumier2017). The introduction of avian haemosporidian parasites to Hawaii led to widespread declines and likely extinctions of forest birds, demonstrating the potential of these parasites to become emerging infectious diseases in naïve host populations (van Riper et al., Reference van Riper, Sandra, van Riper, Lee and Marshal1986; Beadell et al., Reference Beadell, Ishtiaq, Covas, Melo, Warren, Atkinson, Bensch, Graves, Jhala, Peirce, Rahmani, Fonseca and Fleischer2006). Genetic barcoding methods have revealed thousands of avian haemosporidian genetic lineages (Bensch et al., Reference Bensch, Hellgren and Pérez-Tris2009) which, in many cases, seem to constitute true biological species (Bensch et al., Reference Bensch, Pérez-Tris, Waldenström and Hellgren2004; Outlaw and Ricklefs, Reference Outlaw and Ricklefs2014; Hellgren et al., Reference Hellgren, Atkinson, Bensch, Albayrak, Dimitrov, Ewen, Kim, Lima, Martin, Palinauskas, Ricklefs, Sehgal, Valkiūnas, Tsuda and Marzal2015).
The geographic distributions of haemosporidian parasites have been investigated in the context of the migration patterns of their avian hosts. For example, in a survey of resident and migratory birds at a stop-over site in Nigeria, Waldenström et al. (Reference Waldenström, Bensch, Kiboi, Hasselquist and Ottosson2002) identified phylogenetic clades of parasite lineages that might have been transmitted either in Europe, in Africa or in both continents. A similar study of avian haemosporidian parasites of birds in the Americas found similar regional specialization but less phylogenetic signal in parasite transmission areas (Ricklefs et al., Reference Ricklefs, Medeiros, Ellis, Svensson-Coelho, Blake, Loiselle, Soares, Fecchio, Outlaw, Marra, Latta, Valkiūnas, Hellgren and Bensch2017). Indeed, differences in distributions among haemosporidian genera have been linked to the probability that the parasites infect migratory host species (Hellgren et al., Reference Hellgren, Waldenström, Peréz-Tris, Ösi, Hasselquist, Krizanauskiene, Ottosson and Bensch2007). However, some avian haemosporidian parasites, such as the lineage GRW04 (morphological species Plasmodium relictum) – the parasite that contributed to the decline of Hawaii's avifauna (Beadell et al., Reference Beadell, Ishtiaq, Covas, Melo, Warren, Atkinson, Bensch, Graves, Jhala, Peirce, Rahmani, Fonseca and Fleischer2006) – have nearly global distributions that cannot be attributed to natural avian migration patterns and might be related to human introductions of birds and vectors to novel areas (Soares et al., Reference Soares, Marra, Gray and Ricklefs2017). Still other haemosporidians may have dispersed large distances without the help of humans and outside of typical bird migration routes. For example, Ricklefs and Fallon (Reference Ricklefs and Fallon2002) explicitly investigated the regional distributions of avian haemosporidian parasites and found many examples of related lineages restricted to more than one biogeographic region. Similarly, Beadell et al. (Reference Beadell, Ishtiaq, Covas, Melo, Warren, Atkinson, Bensch, Graves, Jhala, Peirce, Rahmani, Fonseca and Fleischer2006) identified an African and a Eurasian parasite nested within a well-supported clade of mostly New World parasite lineages, suggesting long-distance, overwater dispersal events in the evolutionary history of those parasites. Additional patterns likely representing long-distance dispersal events can be identified from other avian haemosporidian phylogenies (e.g. Ewen et al., Reference Ewen, Bensch, Blackburn, Bonneaud, Brown, Cassey, Clarke and Pérez-Tris2012; Cornuault et al., Reference Cornuault, Warren, Bertrand, Milá, Thébaud and Heeb2013; Murdock et al., Reference Murdock, Adler, Frank and Perkins2015).
The mechanisms by which long-distance dispersal of avian haemosporidian parasites may have occurred (i.e. via avian or dipteran hosts, natural or human-mediated) are unclear. An analysis of the molecular substitution rate of extant haemosporidian parasites indicates that these parasites may have diversified across mammals, birds and reptiles within the last 20 million years, and across birds and reptiles in the last 10 million years (Ricklefs and Outlaw, Reference Ricklefs and Outlaw2010), although this estimate is controversial (Bensch et al., Reference Bensch, Hellgren, Križanauskienė, Palinauskas, Valkiūnas, Outlaw and Ricklefs2013; Pacheco et al., Reference Pacheco, Matta, Valkiūnas, Parker, Mello, Stanley, Lentino, Garcia-Amado, Cranfield, Kosakovsky Pond and Escalante2018). However, regardless of the rate of haemosporidian evolution, frequent overwater, long-distance dispersal events are indicated by the present-day global distribution of most major clades. Current parasite distributions reveal many examples of closely related haemosporidian lineages occupying disparate geographic localities, and several lineages appear to have nearly global distributions, providing the strongest evidence of long-distance dispersal.
Here, we investigated the phylogenetic relationships, historical biogeography, and current distributions of presently known avian haemosporidian parasites. Our goal was to place the geographic distributions of these parasites in an evolutionary context and to quantify global patterns of dispersal. Understanding the global biogeography of these parasites is an important step in determining the likelihood that lineages may disperse between regions and become emerging infectious diseases in the future.
Materials and methods
Data
We downloaded a 479 bp fragment of the cytochrome b gene from all parasite lineages in the MalAvi database (http://mbio-serv2.mbioekol.lu.se/Malavi/; Bensch et al., Reference Bensch, Hellgren and Pérez-Tris2009) belonging to the genera Plasmodium, Haemoproteus and Leucocytozoon on 13 September 2017 (MalAvi v.2.3.2). Several of the MalAvi sequences probably represent variations on a single lineage because they differed by one or a few ambiguous base pairs only. We randomly selected one sequence from each repeated haplotype and removed the others using the ‘clean_alignment’ function in the R package malaviR (Ellis et al., Reference Ellis, Bensch and Canbäck2017). We then compared this group of parasite lineages with lineages defined in several published studies from RE Ricklefs' laboratory (e.g. Ricklefs et al., Reference Ricklefs, Outlaw, Svensson-Coelho, Medeiros, Ellis and Latta2014; Ellis et al., Reference Ellis, Collins, Medeiros, Sari, Coffey, Dickerson, Lugarini, Stratford, Henry, Merrill, Matthews, Hanson, Roberts, Joyce, Kunkel and Ricklefs2015) and one unpublished survey (Table S1). We only compared sequences with complete or nearly complete overlap in the 479 bp gene fragment. We found 62 lineages that were identical among the MalAvi and Ricklefs databases (Table S2), and for those we retained the MalAvi names. We identified an additional 47 lineages that were not found in MalAvi (Table S3); all new lineages and alternative lineages names have been submitted to MalAvi.
We then added geographic information to lineages in the MalAvi database that were missing information by consulting original publications (Table S4) and using the Ricklefs database (Table S5). We identified geographic information for 861 Plasmodium lineages (out of 875 in the full alignment), 938 Haemoproteus lineages (out of 950) and 653 Leucocytozoon lineages (out of 673). Lineages were assigned to one or more of 11 biogeographic regions (Europe, North Africa and the Middle East, South Sahara, Asia, Oceania, Australia and New Zealand, North America, Central America, South America, Hawaii and Antarctica; Supplementary Fig. S1) following the current classifications in MalAvi.
Phylogenetic analyses
For each parasite genus, we created a maximum likelihood phylogenetic tree based on the cytochrome b sequences using the software RAxML Blackbox (Stamatakis et al., Reference Stamatakis, Hoover and Rougemont2008) with a γ model of rate heterogeneity and bootstrap support values for the nodes (Supplementary Figs S2–S4). Most nodes in these trees were unresolved, so we identified nodes with bootstrap support ⩾0.8 and with geographic information for all descendant lineages. We excluded one deep node in the Haemoproteus phylogeny that had 52 descendant lineages and multiple biogeographic regions on either side of the node making it difficult to classify as allopatric or sympatric (see below), and different than the rest of the nodes that were mostly shallow on the phylogeny. This left for the analysis 100 nodes for Plasmodium, 142 nodes for Haemoproteus and 96 nodes for Leucocytozoon (Tables S6 and S7), the majority of which were shallow (mean number of descendant lineages per node for Plasmodium was 3.94 ± 0.45 s.e.; Haemoproteus, 3.85 ± 0.30; Leucocytozoon, 3.82 ± 0.47). We scored these nodes as having sister lineages distributed in sympatry or in allopatry (Supplementary Fig. S5). For nodes with three or more descendant lineages, we conservatively classified sister lineages as being in sympatry if at least one lineage on one side of a node was found in the same geographic area as any lineage on the other side of the node. Such coarse-grain geographic classifications might misrepresent some short-distance allopatric events as sympatric events. However, we were primarily interested in identifying long-distance allopatric dispersal and we expected that uneven sampling across smaller geographic scales and across studies could produce false short-distance allopatric events. We therefore chose the coarse-grain classifications in MalAvi, while acknowledging the shortcomings of this approach. Finally, we determined whether allopatric nodes connected biogeographic regions that fall along three prominent avian migratory flyways, the Nearctic–Neotropical flyway, the Palaearctic–Afrotropical flyway and the Asian–Australasian flyway (Newton, Reference Newton2007).
Molecular variation among biogeographic regions
We were also interested in how much of the molecular variation among parasite lineages could be explained by biogeographic regions. Accordingly, we restricted the lineages to those occurring in only one biogeographic region (the vast majority of lineages; see Fig. 1) and calculated the raw genetic distances among those lineages (separately for each genus) based on their cytochrome b sequences using the ‘dist.dna’ function in the R package ape (Paradis et al., Reference Paradis, Claude and Strimmer2004). We then tested whether biogeographic region could explain any of the variation in the genetic distance matrices using the ‘adonis’ function in the R package vegan (Oksanen et al., Reference Oksanen, Blanchet, Friendly, Kindt, Legendre, McGlinn, Minchin, O'Hara, Simpson, Solymos, Stevens, Szoecs and Wagner2017), which performs a permutational analysis of variance (ANOVA) on distance matrices equivalent to an ‘analysis of molecular variance’ (Excoffier et al., Reference Excoffier, Smouse and Quattro1992). We ran 999 permutations per test. All analyses were run with R v.3.4.0 (R Core Team, 2017) and graphics were produced using the R packages ggplot2 (Wickham, Reference Wickham2009) and phytools (Revell, Reference Revell2012).
Frequency distributions of the number of biogeographic regions (11 total regions) occupied by parasite lineages in each genus. The majority of lineages were found in a single region, but several, especially in Plasmodium, were globally distributed (see list of lineages distributed in more than one region in Table S8).
Results
Although the majority of the lineages in each genus were recovered from hosts in a single biogeographic region (Fig. 1), some lineages were globally distributed (Table S8; however, no lineages were found in Antarctica). For example, the Plasmodium lineage GRW04 was found in all regions except Antarctica, and the Plasmodium lineage GRW06 was found in eight regions (only missing from Hawaii, Central America and Antarctica). The most broadly distributed Haemoproteus lineage (PYERY01) was recovered from birds in six biogeographic regions (Europe, North Africa and the Middle East, South Sahara, North America, Central America and South America) and the most broadly distributed Leucocytozoon lineage (SFC8) occurred in four regions (Europe, North Africa and the Middle East, South Sahara and Asia). Overall, 12% of Plasmodium lineages (100/861), 15% of Haemoproteus lineages (139/938) and 7% of Leucocytozoon lineages (47/653) were distributed in more than one biogeographic region.
Among the well-supported nodes in the phylogenetic analysis, most descendant sister lineages were distributed in the same biogeographic region (89/100 Plasmodium nodes, 119/142 Haemoproteus nodes and 82/96 Leucocytozoon nodes were sympatric). Although these sympatric nodes were heterogeneously distributed among biogeographic regions (likely representing heterogeneous sampling), the distributions differed among the parasite genera (χ 2 = 76.3, d.f. = 16, P < 0.001; Fig. 2). Specifically, sympatric nodes in the Plasmodium phylogeny were most often found connecting lineages in South America (χ 2 = 122.3, d.f. = 8, P < 0.001), whereas sympatric Haemoproteus nodes were most often found connecting lineages in Europe, South Sahara and South America (χ 2 = 39.3, d.f. = 8, P < 0.001). Additionally, sympatric nodes in the Leucocytozoon phylogeny were most often found connecting lineages in Asia, North America, Europe and South Sahara (χ 2 = 59.9, d.f. = 8, P < 0.001). Neither the Plasmodium nor the Leucocytozoon phylogenies had sympatric nodes connecting lineages in Central America. In most cases, sympatric nodes connected descendant sister lineages occupying a single region. However, some sympatric nodes connected sister lineages that occur in more than one region (two Plasmodium nodes and 11 Haemoproteus nodes; Tables S6 and S7). For example, node 1030 in the Plasmodium phylogeny connects two lineages, PESA01 and NYCNYC01, both of which have been reported from both North America and South America.
Relative proportion of sympatric nodes connecting sister lineages by biogeographic region, calculated separately for each parasite genus (i.e. for each parasite genus, the number of nodes connecting sister lineages distributed in a particular biogeographic region divided by the total number of sympatric nodes identified). Sympatric nodes connecting sister lineages in more than one biogeographic region were excluded from this analysis.
A substantial proportion of the nodes connected descendant sister lineages distributed in allopatry, i.e. different continental regions; 11% of Plasmodium nodes (11/100), 16% of Haemoproteus nodes (23/142) and 15% of Leucocytozoon nodes (14/96). Many of these allopatric sister lineages were separated by considerable distances. As expected, many allopatric sister lineages were distributed in regions that fall along known avian migratory flyways, however some were not (Table S9, Figs 3 and 4).
Examples of allopatric nodes from the Plasmodium phylogeny. The top map shows two lineages distributed in Europe that are sister to three lineages distributed in Asia and South Sahara. The bottom map shows an allopatric node connecting two lineages on opposite sides of the Atlantic Ocean (North America and South Sahara).
Allopatric nodes (i.e. those connecting sister lineages distributed in different biogeographic regions) classified by whether they fall along three prominent avian migratory flyways (A–A, Asian–Australasian; N–N, Nearctic–Neotropical; P–A, Palaearctic–Afrotropical) or not. The allopatric nodes that do not connect lineages along those migratory flyways (‘none’) represent possible trans-oceanic dispersal events (e.g. across the Atlantic Ocean). Full details regarding these nodes can be found in Table S9.
We also tested how much of the molecular variation among parasite lineages could be explained by biogeographic region using ANOVA for distance matrices. Biogeographic region explained a small but significant fraction of the variation in genetic distances among lineages for all three parasite genera (Plasmodium F 8,752 = 3.76, R 2 = 0.04, P = 0.001; Haemoproteus F 8,790 = 5.72, R 2 = 0.05, P = 0.001; Leucocytozoon F 6,599 = 5.54, R 2 = 0.05, P = 0.001). However, the vast majority of the variation remained unexplained by region, emphasizing the global-scale homogenization of haemosporidian lineages by long-distance dispersal.
Discussion
Dispersal of avian haemosporidian parasites appears to have been relatively unimpeded by geographic barriers on a global scale. We found parasite lineages distributed across multiple continents, only weak genetic structure across parasite distributions, and evidence suggesting that 10–15% of speciation events occurred in allopatry with respect to major biogeographic regions. The distributions of several allopatric sister lineages did not fall along regular avian migratory flyways (Table S9, Figs 3 and 4), and it is unlikely that migratory birds could have fully shaped the distributions of the globally distributed lineages. A high dispersal capacity of haemosporidians has been implicit in the literature since avian infections began to be identified molecularly. In an early study, Waldenström et al. (Reference Waldenström, Bensch, Kiboi, Hasselquist and Ottosson2002) identified sister clades of avian haemosporidian parasites distributed allopatrically in Europe and Asia, suggesting historical parasite dispersal events between the continents. At the same time, Ricklefs and Fallon (Reference Ricklefs and Fallon2002) explicitly investigated the distributions of parasite sister lineages, finding that more diverged lineages were more often found distributed in allopatry and on different host species than less diverged lineages. Hellgren et al. (Reference Hellgren, Waldenström, Peréz-Tris, Ösi, Hasselquist, Krizanauskiene, Ottosson and Bensch2007) presented further evidence of this pattern, and suggested that it may be more common in Plasmodium than in Haemoproteus. Furthermore, several studies have documented global distributions of some haemosporidian parasite lineages (e.g. Beadell et al., Reference Beadell, Ishtiaq, Covas, Melo, Warren, Atkinson, Bensch, Graves, Jhala, Peirce, Rahmani, Fonseca and Fleischer2006; Hellgren et al., Reference Hellgren, Atkinson, Bensch, Albayrak, Dimitrov, Ewen, Kim, Lima, Martin, Palinauskas, Ricklefs, Sehgal, Valkiūnas, Tsuda and Marzal2015).
Beyond confirming the potential for haemosporidian parasites to become emerging infectious diseases in the future (and perhaps having been emerging infectious diseases many times in the past), the frequency of long-distance dispersal of haemosporidians has implications for dating the haemosporidian phylogeny using biogeographic calibration points. For example, a monophyletic clade of Plasmodium parasites is known to infect the lemurs of Madagascar (Pacheco et al., Reference Pacheco, Battistuzzi, Junge, Cornejo, Williams, Landau, Rabetafika, Snounou, Jones-Engel and Escalante2011). This fact led Pacheco et al. (Reference Pacheco, Battistuzzi, Junge, Cornejo, Williams, Landau, Rabetafika, Snounou, Jones-Engel and Escalante2011) to suggest that the last date of terrestrial mammalian colonization of Madagascar (ca. 20 million years ago) could be used as the minimum age of lemur Plasmodium parasite divergence. However, others have pointed out that the ancestor of lemur Plasmodium could have arrived more recently via mosquito dispersal since vectors may be able to disperse parasites relatively long distances (Bensch et al., Reference Bensch, Hellgren, Križanauskienė, Palinauskas, Valkiūnas, Outlaw and Ricklefs2013). If vectors were able to move the parasites to Madagascar at any time, then divergence of lemur Plasmodium from their ancestor could also have happened at any time, rendering that node in the haemosporidian phylogeny impossible to date. Our analysis supports the plausibility of such a scenario (mosquito-mediated dispersal of Plasmodium) and casts doubt on the use of that biogeographic event (the cessation of mammalian dispersal to Madagascar) as a calibration point in the haemosporidian phylogeny.
We also found that the geographic distributions of sympatric sister lineages differed among the parasite genera. Sympatric lineages of Plasmodium were mostly distributed in South America, while sympatric lineages of Haemoproteus were mostly in Europe, South Sahara, and South America, and sympatric lineages of Leucocytozoon were mostly in Asia, North America, Europe and South Sahara (Fig. 2). With the data available, we could not determine whether this is a result of differential rates of diversification among biogeographic regions or of differential sampling. One potential bias concerns how lineages are defined. Lineages in MalAvi are defined by single base pair differences, while lineages from the Ricklefs database sometimes incorporate variation in the cytochrome b gene. One consequence of this difference in lineage assignment is that the MalAvi lineages are more likely to group into well-supported sister lineages, many of which differ at single nucleotide positions and are more likely to share geographic distributions. Nevertheless, the geographic differences among the parasite genera in a number of sympatric lineages do, to some extent, mirror global patterns of diversity of these parasites (Clark et al., Reference Clark, Clegg and Lima2014). Furthermore, restricting the data to MalAvi lineages and location information results in a higher proportion of nodes connecting allopatrically distributed sister lineages (17.3% of Plasmodium nodes, 21.7% of Haemoproteus nodes and 27.7% of Leucocytozoon nodes), suggesting that the different approaches to defining lineages did not affect our main conclusion of substantial intercontinental dispersal.
Understanding how haemosporidian parasites have dispersed across extensive ocean barriers remains a challenge. We consider three scenarios: (1) the natural movement of avian hosts; (2) the natural movement of dipteran vectors; and (3) within-region diversification followed by recent human-mediated transportation of either avian or dipteran hosts. As stated previously, the first scenario is plausible because birds migrate and often carry their haemosporidian parasites with them (Ricklefs and Fallon, Reference Ricklefs and Fallon2002; Waldenström et al., Reference Waldenström, Bensch, Kiboi, Hasselquist and Ottosson2002; Hellgren et al., Reference Hellgren, Waldenström, Peréz-Tris, Ösi, Hasselquist, Krizanauskiene, Ottosson and Bensch2007; Mendes et al., Reference Mendes, Pardal, Morais, Antunes, Ramos, Perez-Tris and Piersma2013; Ricklefs et al., Reference Ricklefs, Medeiros, Ellis, Svensson-Coelho, Blake, Loiselle, Soares, Fecchio, Outlaw, Marra, Latta, Valkiūnas, Hellgren and Bensch2017). For example, Pérez-Tris and Bensch (Reference Pérez-Tris and Bensch2005) sampled several populations of the avian host Sylvia atricapilla for haemosporidians in the host's breeding and wintering ranges and found that some parasites present in the breeding range also occurred in the blood of birds in their wintering range. Furthermore, some haemosporidian lineages that infect breeding North American birds have also been recovered on the wintering grounds of their hosts from resident birds in the West Indies (Fallon et al., Reference Fallon, Bermingham and Ricklefs2005; Ricklefs et al., Reference Ricklefs, Outlaw, Svensson-Coelho, Medeiros, Ellis and Latta2014). Migrating avian hosts have also been implicated in moving haemosporidian parasites from mainland areas to islands. For example, avian Plasmodium in the Galapagos Islands likely arrived via a migrating host, the Bobolink (Dolichonyx oryzivorus), which regularly stops over in the Galapagos archipelago during its migration (Levin et al., Reference Levin, Zwiers, Deem, Geest, Higashiguchi, Iezhova, JiméNez-UzcáTegui, Kim, Morton, Perlut, Renfrew, Sari, Valkiunas and Parker2013, Reference Levin, Colborn, Kim, Perlut, Renfrew and Parker2016).
Parasite dispersal events over long distances that do not fall along typical avian migration routes are harder to explain by avian host movement. Few avian species regularly cross the Atlantic Ocean as part of their annual migrations (one of the only passerines known to do this is the northern wheatear, Oenanthe oenanthe; Bairlein et al., Reference Bairlein, Norris, Nagel, Bulte, Voigt, Fox, Hussell and Schmaljohann2012). However, several species are regular vagrants between Europe and the eastern USA, and these have been implicated in the expansion of the mosquito-vectored West Nile Virus from the Old to the New World (Rappole et al., Reference Rappole, Derrickson and Hubálek2000). Over evolutionary time, vagrant birds may provide ample opportunity for long-distance parasite dispersal, assuming vagrants are exposed to suitable vectors.
The random, passive movement of dipteran vectors might also result in long-distance dispersal by haemosporidian parasites. Kay and Farrow (Reference Kay and Farrow2000) sampled mosquitos with aerial kite traps up to 310 m in altitude and used prevailing wind speeds to estimate the distance travelled. The authors found that numerous Aedes, Anopheles and Culex mosquitos had travelled an average of 152 km and, accordingly, could have introduced a viral pathogen to Australia. Furthermore, Asahina (Reference Asahina1970) collected several mosquito species 500 km at sea, off the coast of Japan. Culicoides biting midges have also been found to travel hundreds of kilometres in wind currents, even over large bodies of water (Burgin et al., Reference Burgin, Gloster, Sanders, Mellor, Gubbins and Carpenter2013). Pettersson et al. (Reference Pettersson, Bensch, Ander, Chirico, Sigvald and Ignell2013) captured ornithophilic Culicoides biting midges 12 m above the ground in southern Sweden, further suggesting that they may be carried long distances by winds. These results suggest that dipteran vectors could potentially facilitate haemosporidian dispersal over long distances.
A final dispersal scenario involves humans transporting parasites outside of the regions in which they diversified. This has likely happened in the past. The avian malaria parasite GRW04 probably arrived to the Hawaiian Islands with introduced birds and was able to spread due to the presence of the introduced mosquito vector Culex pipiens quinquefasciatus (Beadell et al., Reference Beadell, Ishtiaq, Covas, Melo, Warren, Atkinson, Bensch, Graves, Jhala, Peirce, Rahmani, Fonseca and Fleischer2006). Furthermore, Ewen et al. (Reference Ewen, Bensch, Blackburn, Bonneaud, Brown, Cassey, Clarke and Pérez-Tris2012) demonstrated that the introduction of haemosporidian parasites to New Zealand likely occurred as a result of the introduction of infected birds from Europe.
Many researchers have documented the geographic spread and host-switching of blood-borne pathogens (Escalante et al., Reference Escalante, Cornejo, Freeland, Poe, Durrego, Collins and Lal2005; Kilpatrick et al., Reference Kilpatrick, Chmura, Gibbons, Fleischer, Marra and Daszak2006), and it is clear that they are often capable of dispersing long distances and switching across large numbers of host species (Peterson, Reference Peterson2008). This also seems to be the case with avian haemosporidian parasites (Ricklefs et al., Reference Ricklefs, Outlaw, Svensson-Coelho, Medeiros, Ellis and Latta2014), which appear to have undergone multiple long-distance dispersal events over their evolutionary history. Future research might address haemosporidian dispersal by (1) screening migrating and vagrant avian hosts, (2) documenting vector dispersal patterns and the potential effect of haemosporidian infection on vector movement (Levin and Parker, Reference Levin and Parker2014), and (3) increasing regional-level sampling of haemosporidian parasites to boost confidence in parasite distributions. Furthermore, as more genetic data are produced, phylogenies should become better resolved (Bensch et al., Reference Bensch, Canbäck, DeBarry, Johansson, Hellgren, Kissinger, Palinauskas, Videvall and Valkiūnas2016; Videvall et al., Reference Videvall, Cornwallis, Ahrén, Palinauskas, Valkiūnas and Hellgren2017), affording researchers the opportunity to test whether long-distance dispersal events influence the rates of diversification. While most studies of parasite dispersal are undertaken to identify the origins of emerging infectious diseases, additional studies might address the dispersal patterns of entire parasite taxa to better understand the historical context that allows parasites to arise in new areas and infect naïve hosts.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0031182018001130.
Acknowledgements
Discussions with Matthew Medeiros, Alan Fecchio, and members of the Molecular Ecology and Evolution Lab at Lund University and comments from two anonymous reviewers improved the manuscript.
Financial support
We thank the US National Science Foundation and the National Geographic Society for support of fieldwork, and the Curators of the University of Missouri and the Whitney Harris World Ecology Center for funding laboratory work. D.R.R. was supported by the US National Science Foundation (IOS-1121435 and IOS-1257530). V.A.E. was supported by a postdoctoral fellowship from the Carl Tryggers Foundation.
Conflict of interest
None.
Ethical standards
We acknowledge Kenya's National Commission for Science, Technology and Innovation, Kenya's National Environmental Management Authority, the Kenya Wildlife Service and the Mpala Research Centre for enabling this work. This work was approved by Columbia University's IACUC.
