Genetica (2014) 142:235 249 DOI 10.1007/s10709-014-9770-9 Distribution, prevalence and host specificity of avian malaria parasites across the breeding range of the migratory lark sparrow (Chondestes grammacus) Bethany L. Swanson Amanda C. Lyons Juan L. Bouzat Received: 29 May 2013 / Accepted: 22 May 2014 / Published online: 1 June 2014 Ó Springer International Publishing Switzerland 2014 Abstract The lark sparrow (Chondestes grammacus) isa ground-nesting passerine that breeds across much of the central North American steppe and sand barrens. Through genotyping and sequencing of avian malaria parasites we examined levels of malaria prevalence and determined the distribution of Haemoproteus and Plasmodium lineages across the breeding range of the lark sparrow. Analysis of 365 birds collected from five breeding locations revealed relatively high levels of malaria prevalence in adults (80 %) and juveniles (46 %), with infections being primarily of Haemoproteus (91 % of sequenced samples). Levels of genetic diversity and genetic structure of malaria parasites with respect to the avian host populations revealed distinct patterns for Haemoproteus and Plasmodium, most likely as a result of their distinct life histories, host specificity, and transmission vectors. With the exception of one common Haemoproteus haplotype detected in all populations, all other haplotypes were either population-specific or shared by two to three populations. A hierarchical analysis of molecular variance of Haemoproteus sequences revealed that 15 18 % of the genetic variation can be explained by differences among host populations/locations (p \ 0.001). In contrast to the regional patterns of genetic differentiation detected for the lark sparrow populations, Haemoproteus parasites showed high levels of population-specific variation and no significant differences among regions, which suggests that the population dynamics of the parasites may be driven by B. L. Swanson A. C. Lyons J. L. Bouzat (&) Department of Biological Sciences, Bowling Green State University, Bowling Green, OH 43403, USA e-mail: jbouzat@bgsu.edu evolutionary processes operating at small spatial scales (e.g., at the level of host populations). These results highlight the potential effects of host population structure on the demographic and evolutionary dynamics of parasites. Keywords Avian malaria Genetic structure Haemoproteus Plasmodium Wildlife disease Introduction Avian malaria parasites are globally distributed among birds, infecting a wide range of host species (Valkiūnas 2005). To date, many studies of avian malaria have not shown direct evidence for a significant negative effect of the disease on the fitness of its avian hosts (e.g. Weatherhead 1990; Dale et al. 1996; Kilpatrick et al. 2006; Knowles et al. 2010; Lachish et al. 2011a, b). However, Bonneaud et al. (2006) showed that specific alleles of the major histocompatibility complex were linked to increased host resistance to malaria in the house sparrow. Furthermore, experimental infections have shown that avian malaria parasites can have significant consequences on their hosts, including enlargement of the liver and spleen, declines in body weight, and even death (Atkinson et al. 1995, 2000). Avian malaria has also been implicated in the severe decline of some bird species endemic to islands. For example, several Hawaiian endemic species have had widespread declines and have even been driven to possible extinction as a result of malaria infections (van Riper et al. 1986; Atkinson et al. 1995, 2000; Massey et al. 1996; Beadell et al. 2006). As avian malaria represents a potential threat to species of conservation concern (see Ishak et al. 2008), it is important to address its potential impact on wild populations.
236 Genetica (2014) 142:235 249 The evolutionary history of parasites and their hosts plays an important role in resistance and susceptibility to infection with malaria. Avian malaria itself is caused by related parasites in two genera, Plasmodium and Haemoproteus, in the phylum Apicomplexa, class Haemosporida (Atkinson and van Riper 1991). The two genera differ in both their primary vectors and in certain stages of their life cycles. The infective stages of both parasites in their vertebrate hosts are differentiated sex cells, which reside inside red blood cells in the peripheral circulation. Haemoproteus undergoes asexual reproduction in various tissues of the vertebrate host prior to entering the blood circulation while Plasmodium undergoes additional cycles of asexual reproduction in the peripheral blood. Both parasite gametocytes undergo sexual fusion and develop sporozoite-containing oocytes inside dipteran primary hosts, which for Plasmodium are mostly Culex or other culicid mosquitoes (Culicidae) and for Haemoproteus mostly Culicoides midges (Ceratopogonidae) (Atkinson and van Riper 1991). The relationships between parasites and their hosts are complex with many interacting factors. Because malaria is a vector-born infection, there is a strong environmental component associated with this disease. In order for a bird to become infected, it must be infected through a vector (e.g., mosquito or midge) containing parasites able to infect that particular host bird species. Thus, the area inhabited by the bird must have suitable vectors as well as an already circulating population of parasites capable of infecting that host species. It should be noted though that host specificity of avian malaria parasites can range from a single host to a wide range of host species that may only be distantly related (Bensch et al. 2000; Ricklefs and Fallon 2002; Waldenstrom et al. 2002; Fallon et al. 2003; Ricklefs et al. 2004; Njabo et al. 2011). In fact, earlier studies have suggested that Haemoproteus are more likely to be host specialists, being species-specific parasites, than Plasmodium (Atkinson and van Riper 1991). Several studies have estimated levels of malaria parasite prevalence in birds (defined simply as the proportion of infected individuals) and have focused on prevalence either in individual species (e.g., Bensch et al. 2007; Wood et al. 2007) or at the community level (e.g., Fallon et al. 2003; Ricklefs et al. 2005). Published infection rates are highly variable for different species. For example, in a 4-year study in southern Missouri, individual species infection rates ranged from 11 % in the blue-winged warbler, Vermivora pinus, to 58 % in the yellow-breasted chat, Icteria virens (Ricklefs et al. 2005). Prevalence has also been shown to vary year-to-year and seasonally at a specific locale (e.g., Ricklefs et al. 2005; Bensch et al. 2007; Cosgrove et al. 2008) and geographically, across independent populations of the same species (e.g., Fallon et al. 2005; Gibb et al. 2005). Variation in malaria parasite prevalence may be due to a variety of factors including the host species exposure to suitable vectors, the density of vectors present in a given habitat, whether the parasite community is composed of host generalists or specialists, and the host s immunological response to the parasites (Pérez-Tris et al. 2005; Bonneaud et al. 2006; Cosgrove et al. 2008; Szöll}osi et al. 2011; Loiseau et al. 2012). While a few recent studies (e.g., Fallon et al. 2006; Durrant et al. 2008; Pagenkopp et al. 2008; Szöll}osi et al. 2011) have examined avian malaria prevalence across the breeding range of single species, there is still a paucity of studies examining breeding populations of migratory species across a wide-ranging geographic area. The present study focuses on the lark sparrow (Chondestes grammacus), a ground-nesting passerine with both sedentary and migratory populations that breeds across much of the central North American steppe and sand barrens (Martin and Parrish 2000). Some sedentary populations are found in the milder climates of Texas and California whereas the migratory populations are spread over a large geographic area, ranging from the Canadian prairies to northern Texas and from Oregon to Ohio (Fig. 1). Eastern populations of the lark sparrow have been described as a separate subspecies (C.g. grammacus, Say, 1823), distinct from populations that breed west of the Mississippi River (C.g. strigatus, Swainson, 1827; Baepler 1968). Furthermore, both morphological and genetic studies suggest that California populations represent a demographically independent unit (Ross and Bouzat 2014), consistent with Oberholser s (1932) original description of the Oregon lark sparrow as a distinct subspecies (C.g. actitus, Oberholser 1932; Miller 1941). The eastern populations have become increasingly fragmented over the past century as population sizes have declined. In the state of Ohio the species is currently listed as state-endangered (Ohio Division of Wildlife), as only a small population persists in northwest Ohio, isolated from any other breeding population by more than 300 km. In this study, we used a genotyping technique, cloning, and sequencing of an avian malaria-specific DNA marker to assess levels of infection and potential genetic structuring of avian malaria in the lark sparrow. Specifically, we addressed three research questions. First, is there variation in avian malaria prevalence across independent breeding populations of the lark sparrow? Secondly, is there any genetic structuring of avian malaria parasites with respect to the lark sparrow breeding populations? Thirdly, are there independent parasite lineages and, if so, what is the distribution of malaria lineages in geographically distinct breeding populations of the lark sparrow? The lark sparrow represents a good model species to study the potential role of host population structure on
Genetica (2014) 142:235 249 237 Fig. 1 Breeding range of the lark sparrow as determined by the Breeding Bird Survey Summer Distribution Map, 1994 2003 (Sauer et al. 2008). Shading on map indicates number of birds counted (scale represented in box). Sampling sites (CA 1, CA 2, NE, TX 1, TX 2, IL, and OH) are located within three breeding regions (west, central, east) of the lark sparrow associated with potential migratory flyways. Pie charts indicate haplotype frequencies for malaria parasites found in each population parasites population dynamics. Looking at differences in parasite prevalence and the composition of the parasite community across lark sparrow breeding populations allowed us to assess the potential role of host dispersal and population isolation on the dynamics of avian malaria. Furthermore, the characterization of malaria parasite lineages in the lark sparrow and their ancestral relationships with malaria lineages from other avian species provided insight into differences in host specificity between Haemoproteus and Plasmodium parasites. This study of a wildlife disease system across the breeding range of a migratory bird species will expand our knowledge of how vector-born parasites may affect populations in the wild and how host distribution may affect parasites population dynamics. Materials and methods Field sampling Breeding populations of the lark sparrow included in this study were selected based on migratory behavior and geographic distribution throughout the species breeding range (Fig. 1). Populations were sampled over three breeding seasons (May July 2005 2007) and include populations in California (Tehama County, 2005 2006 and Mendocino County, 2006), Texas (Tom Green County, 2005 2007 and Kimble County, 2005), Ohio (Lucas and Wood Counties, 2005 2007), Illinois (Jo Daviess and Caroll Counties, 2005 2007), and Nebraska (Keith County, 2007). Due to the proximity of the sampling locations within each state, samples are further classified only by state. Sampled populations were assigned to three breeding regions (west, central, and east; Fig. 1) each represented by morphologically defined subspecies (C.g. actitus, C.g. strigatus, and C.g. grammacus, respectively). In the field, birds were captured by mist-netting. Captured birds were assigned to two age classes. Adults were classified as after-hatch-year (AHY) and juveniles, i.e., birds born that year which had fledged and were captured by mist-net, were classified as hatch-year (HY). Nestlings were disregarded, as most parasite infections may not be detected until an incubation period of about 5 10 days (Valkiūnas 2005). Adults were sexed by examination of the cloacal protuberance, measured (wing chord, tail, tarsus, and bill length, depth and width), weighed, and fitted with a US Fish and Wildlife aluminum leg band and plastic color bands arranged in individual specific patterns. The same was done for juveniles with the exception of sexing, which was not possible through cloacal examination. Less than 100 ll of blood was sampled by puncture of the brachial
238 Genetica (2014) 142:235 249 wing vein and stored in Longmire s lysis buffer (Longmire et al. 1986) for further genetic analysis. All fieldwork was carried out under appropriate collecting permits and animal care and use protocols approved by the Institutional Animal Care and Use Committee at Bowling Green State University (IACUC protocol number 05-006). Molecular analysis In the laboratory, DNA was isolated using a standard saltprecipitation extraction (Sambrook et al. 1989), and extracted DNA was suspended in 19 TE buffer. A total of 365 samples were screened for the presence of malaria parasites in the peripheral blood using the polymerase chain reaction (PCR). These included 50 samples from California, 24 from Nebraska, 144 from Texas, 23 from Illinois, and 124 from Ohio. PCR was performed with primers designed to amplify an approximately 380 base pair region of the mitochondrial cytochrome b (cyt b) gene of avian malaria parasites (primers 543F: 5 0 -AAA AAT ACC CTT CTA TCC AAA TCT-3 0 and 926R: 5 0 -CAT CCA ATC CAT AAT AAA GCA T-3 0 ). The primers were designed from published sequences of the cytochrome b gene for Plasmodium (Escalante et al. 1998) as described in Fallon et al. (2003). These primers have been shown to amplify DNA from both Haemoproteus and Plasmodium species from a wide range of avian hosts worldwide (Ricklefs et al. 2005). PCRs were run in 15 ll volumes that contained the following components in their final concentrations: 19 Promega GoTaq Ò Flexi Buffer (Promega Corporation, Madison, WI), 2.5 mm MgCl 2, 0.2 mm of each dntp, 0.5 lm of each primer, 0.375 units of Promega GoTaq Ò DNA polymerase and approximately 50 ng DNA as template. PCR amplification profiles included an initial denaturation for 1 min at 94 C, followed by 35 cycles with denaturation for 50 s at 94 C, annealing for 50 s at 49 C, and extension for 70 s at 72 C, followed by a final extension step of 3 min at 72 C. In order to enhance DNA typing accuracy, at least two negative controls and a positive control were included with each PCR. To attempt to eliminate false negatives, all samples initially screened as negative were subjected to a second PCR amplification. PCR products were checked on a 1 % agarose gel and infection status was determined by the presence of a band at approximately 380 bp. In addition to the malaria typing, each sample was subjected to a PCR designed for the genetic sexing of individuals, following the procedures described in Griffiths et al. (1998). These reactions revealed 95.2 % accuracy of the sexing of adults by cloacal examination. Furthermore, the successful amplification of all samples (as well as previous microsatellite and mtdna amplification; Ross and Bouzat 2014) served as a control for the negative typing of malaria infections, i.e. eliminating the possibility that negative amplifications were associated with low quality of DNA extractions (e.g., due to the presence of PCR inhibitors). To characterize parasite lineages, PCR products were subjected to an ethanol precipitation and resuspended in sterile deionized water for direct sequencing. We initially direct sequenced all individuals that tested positive (255 samples) on an ABI377 automated gene sequencer (Applied Biosystems, Foster City, CA) according to the manufacturer s protocol using Big Dye version 3.1 dyeterminator sequencing chemistry (Applied Biosystems, Foster City, CA). In several occasions, chromatograms showed double peaks at multiple sites evidencing multiple malaria infections (17 samples). When possible, individual haplotypes from double infections were inferred from known haplotype sequences. The remaining samples were subjected to cloning using Promega pgem-t Easy (Promega; Madison, Wisconsin) kits. Two to four clones from each individual were sequenced by the University of Chicago DNA Sequencing Facility using fluorescent dye terminators and standard vector primers. Each clone was sequenced at least twice to confirm the validity of the resulting sequence as a distinct malaria haplotype. The chromatograms from a small number of samples revealed multiple ambiguities that could not be resolved and thus were discarded from analysis (27/255 infected samples had no haplotype assignments). Overall, 89.4 % of individuals found to be infected (228/255 infected samples) were assigned a specific malaria haplotype. Statistical and phylogenetic analysis Samples were grouped into different arrangements (i.e., by sampling location, by region, by age of birds, and by sex) and prevalence was estimated as the number of infected birds (as determined by positive PCR amplification) divided by the total number of individuals sampled. Chi square tests were performed to test for heterogeneity in malaria prevalence among defined groups. Mitochondrial cytochrome b sequences were manually edited by checking and correcting for miscalls and ambiguities. Sequences were aligned using Clustal X multiple sequence alignment software (Larkin et al. 2007) and ambiguous ends were trimmed. This resulted in a 302 bp fragment for use in phylogenetic analyses. Sequences differing by as little as one nucleotide were considered distinct haplotypes. Individual sequences were identified to the genus level, as either Haemoproteus or Plasmodium, by comparison with parasite sequences that had been previously classified. The characterized haplotypes could represent either the same or different species/lineages. Each
Genetica (2014) 142:235 249 239 haplotype was also subjected to a BLAST search to determine if it had previously been found and reported in avian species other than the lark sparrow. Levels of genetic variation were computed in MEGA, version 4 (Tamura et al. 2007). First, genetic variation was computed for Haemoproteus and Plasmodium haplotypes separately based on the number of base pair differences and p-distance among DNA sequences and included estimates of standard errors based on 500 bootstrap replicates. Average values were computed for both within and between the genera. Genetic variation was also computed by region, both within east, central, and west breeding regions, and between paired regions (east central, east west, and central west). To visualize mutational pathways, a haplotype network based on statistical parsimony was generated using TCS 1.2.1 software (Clement et al. 2000). The confidence limit for creating connections between haplotypes was set to 95 %. Estimates of haplotype (h) and nucleotide diversity (p) for Haemoproteus and Plasmodium and their standard deviations were obtained for each host population and all samples in general using Arlequin version 3.0 (Excoffier et al. 2005). Arlequin was also used to perform a hierarchical analysis of molecular variance (AMOVA) and assess potential genetic structure of avian malaria parasites across the lark sparrow breeding range. Genotyped host populations were grouped by breeding region (i.e., west, central and east). Two independent AMOVA were performed based on pairwise differences between haplotypes and haplotype frequencies, respectively. Fixation indices (F CT,F SC,andF ST )wereused to partition the percent of variation explained by differences among regions, between populations within regions, and among individuals within populations, respectively. Significant values for each index were calculated using 1,000 random permutations of haplotypes between populations. Given the paucity of samples that tested positive for Plasmodium, the AMOVA was restricted to Haemoproteus. Pair-wise exact tests of population differentiation were also performed through permutation procedures implemented in Arlequin (10,000 permutations). To evaluate relationships between parasite lineages, two phylogenies were constructed using Bayesian inference (Felsenstein 2004). The first included all of the avian malaria haplotypes found in the lark sparrow and a selected sample of Haemoproteus and Plasmodium parasites found in other avian species. This phylogeny was aimed at determining if malaria parasites from the lark sparrow represented species-specific lineages that have evolved as independent monophyletic groups within the species. The second phylogenetic analysis was based only on the lark sparrow haplotypes characterized in this study and focused on determining potential phylogeographic patterns in relation to the presence/absence of individual parasite lineages in distinct lark sparrow host populations/regions. The program jmodeltest (Posada 2008) was used in order to determine the best-fit model of nucleotide substitution for the data. Seven substitution schemes were utilized for a total of 56 possible models. After computation of likelihood scores for each model, the Akaike Information Criterion (AIC) was used to select the best model for the given sequence data (GTR?G for both analyses). Phylogenetic trees of parasite lineages were constructed using Bayesian Markov chain Monte Carlo (MCMC) analysis in the program BEAST (Drummond and Rambaut 2007). Input files were generated using BEAUTi (included with the BEAST software package). For the MCMC analysis the length of the chain was set to 10,000,000. TreeAnnotator (also part of the BEAST software package) was used to summarize the trees produced in BEAST. Results Across its breeding range, the lark sparrow populations showed relatively high levels of infection with malaria parasites. Overall, 255 out of 365 samples (69.9 %) including juveniles and adults screened positive by PCR. Overall prevalence in adults (AHY birds) was 80.3 % (204 out of 254 samples) and in the HY juveniles (in this study meaning HY birds that had fledged and were captured by mist-net) prevalence was 45.9 % (51 out of 111 samples) (Table 1). Infection rates between adult and juvenile birds differed significantly (v 2 = 43.34, df = 1, p \ 0.001); therefore they will be further considered separately. Considering adult lark sparrows, infection rates in each studied population were as follows: 80.0 % in California (20 of 25 samples), 87.5 % in Nebraska (21 of 24 samples), 80.8 % in Texas (84 of 104 samples), 73.9 % in Illinois (17 of 23 samples), and 79.5 % in Ohio (62 of 78 samples) (see Table 1). There were no significant differences in the rate of infection among populations (v 2 = 1.42, df = 3, p [ 0.5). Furthermore, in populations that were sampled in all 3 years from 2005 to 2007 (i.e., Illinois, Texas, and Ohio), there were no significant differences in infection rates among sampling years (v 2 = 5.51, df = 2, p [ 0.05). There was, however, a significant difference in malaria prevalence when the sex of adult birds was considered (v 2 = 5.94 df = 1, p \ 0.05), with females having a significantly higher infection rate (87.3 %) than males (75.0 %). In populations where a substantial number of juvenile birds were sampled, infection rates of juveniles were consistently lower than those detected in adults of the same population (Fig. 2). Malaria prevalence in juveniles was 28.0 % (7 of 25 samples) in California, 67.5 % (27 of 40 samples) in Texas, and 37.0 % (17 of 46 samples) in Ohio. In contrast to that observed in adults, differences in the rate
240 Genetica (2014) 142:235 249 Table 1 Avian malaria prevalence in the lark sparrow estimated by population (identified by state abbreviations: CA, NE, TX, IL, OH) and by age class (after hatch-year, AHY; hatch-year, HY) Population Age class N Ninf Prevalence Nseq Haem Plas Individuals Double infec CA AHY 25 20 0.80 19 0 16 3 HY 25 7 0.28 6 0 6 0 NE AHY 24 21 0.88 20 0 19 1 HY TX AHY 104 84 0.81 76 11 85 2 HY 40 27 0.68 24 2 26 0 IL AHY 23 17 0.74 17 1 17 1 HY OH AHY 78 62 0.79 52 3 53 2 HY 46 17 0.37 14 0 1 13 Overall AHY 254 204 0.80 184 15 190 9 HY 111 51 0.46 44 2 33 13 Total 365 255 0.70 228 17 223 22 N refers to sample size, Ninf refers to number of infected samples, Nseq identifies the number of individuals sequenced in each population (individuals) and the number of individuals with double infections (double infec) found in each population. Sequences are reported independently for Haemoproteus (Haem) and Plasmodium (Plas) Fig. 2 Avian malaria prevalence in adult and juvenile lark sparrows. Significant differences between juvenile and adult malaria prevalence were found in the California and Ohio populations, with adults having a significantly higher rate of infection (***p \ 0.05) of infection of juveniles among populations were significant (v 2 = 12.22, df = 2, p \ 0.005). Furthermore, differences between adult and juvenile infection rates were found to be significant in California (v 2 = 13.61, df = 1, p \ 0.005) and Ohio (v 2 = 22.64, df = 1, p \ 0.005), but not in Texas (v 2 = 2.66, df = 1, p [ 0.1). Malaria prevalence was also evaluated across eastern, central and western breeding regions of the lark sparrow. There were no significant differences in the rate of infection by region (v 2 = 0.52, df = 2, p [ 0.5). Each region was also evaluated for differences in prevalence by both age and sex of the birds. In the eastern and western regions, adults were found to have a significantly higher infection rates than juvenile birds (eastern, v 2 = 23.75, df = 1, p \ 0.001 and western, v 2 = 13.61, df = 1, p \ 0.001). In the central region, however, there was no significant difference by age class (v 2 = 3.72, df = 1, p [ 0.05). Differences in adult prevalence by sex also showed variation among regions. In the central region, adult females were found to have a significantly higher rate of infection than males (v 2 = 6.10, df = 1, p \ 0.05) while in the eastern and western regions, no significant differences were found (east, v 2 = 0.62, df = 1, p [ 0.25 and west, v 2 = 1.04, df = 1, p [ 0.25). Overall, 89.4 % of individuals found to be infected (228/255 infected samples) were assigned a specific malaria haplotype. A total of 228 of the 255 samples (89.4 %) that tested positive for malaria infections were successfully sequenced to obtain the parasite cyt b mtdna sequences. These samples resulted in a total of 245 distinct sequences. Of those, 91 % (223 haplotypes) were identified as Haemoproteus sequences while only 9 % (22 haplotypes) were identified as Plasmodium (Table 1). A number of individuals (17) revealed double infections (i.e., sequencing indicated the presence of two distinct haplotypes). Most double infections revealed haplotypes from either Haemoproteus or Plasmodium, although in two occasions, individuals with mixed infections including Haemoproteus and Plasmodium were identified. Haemoproteus sequences consisted of 14 distinct haplotypes
Genetica (2014) 142:235 249 241 Table 2 Lark sparrow malaria haplotypes with BLAST matches found in GenBank GenBank accession number (GenBank #), host species in which the matching parasite sequence was found, host family, and reference study (source) are reported Lineage GenBank # Host species Host family Source H1 KJ735820 Chondestes grammacus Emberizidae This study H2 KJ735821 Chondestes grammacus Emberizidae This study H3 KJ735822 Chondestes grammacus Emberizidae This study AF465562 Sialia mexicana, S. sialis Turdidae Ricklefs and Fallon (2002) H4 KJ735823 Chondestes grammacus Emberizidae This study H5 KJ735824 Chondestes grammacus Emberizidae This study HU1 KJ735825 Chondestes grammacus Emberizidae This study AF465580 Dendroica pensylvanica Parulidae Ricklefs and Fallon (2002) HU2 KJ735826 Chondestes grammacus Emberizidae This study HU3 KJ735827 Chondestes grammacus Emberizidae This study HU4 KJ735828 Chondestes grammacus Emberizidae This study HU5 KJ735829 Chondestes grammacus Emberizidae This study HU6 KJ735830 Chondestes grammacus Emberizidae This study HU7 KJ735831 Chondestes grammacus Emberizidae This study HU8 KJ735832 Chondestes grammacus Emberizidae This study HU9 KJ735833 Chondestes grammacus Emberizidae This study P1 KJ735813 Chondestes grammacus Emberizidae This study DQ659541 Carpodacus mexicanus Fringillidae Beadell et al. (2006) DQ659539 Molothrus ater Icteridae Beadell et al. (2006) P2 KJ735814 Chondestes grammacus Emberizidae This study AF465556 Passerina cyanea Cardinalidae Ricklefs and Fallon (2002) AY733088 Spheniscus demersus Spheniscidae Beadell and Fleischer (2005) AY167247 Coereba flaveloa Coerebidae Fallon et al. (2003) EU627843 Strix varia Strigidae Ishak et al. (2008) JN164732 Sylvia atricapilla Sylviidae Direct submission GQ141569 Turdus fumigatus Turdidae Outlaw and Ricklefs (2009) GQ141570 Geotrygon montana Columbidae Outlaw and Ricklefs (2009) GQ141574 Mniotilta varia Parulidae Outlaw and Ricklefs (2009) P3 KJ735815 Chondestes grammacus Emberizidae This study AY540210 Dendroica coronata Parulidae Ricklefs et al. (2004) HM222474 Catharus ustulatus Turdidae Ricklefs and Outlaw (2010) P4 KJ735816 Chondestes grammacus Emberizidae This study AY540222 Vireo solitarius Vireonidae Ricklefs et al. (2004) PU1 KJ735817 Chondestes grammacus Emberizidae This study PU2 KJ735818 Chondestes grammacus Emberizidae This study PU3 KJ735819 Chondestes grammacus Emberizidae This study JN819356 Turdus assimilis Turdidae Direct submission HM222480 Hylocichla mustelina Turdidae Ricklefs and Outlaw (2010) HM222477 Turdus migratorius Turdidae Ricklefs and Outlaw (2010) (GenBank accession numbers KJ735820 KJ735833), with two common haplotypes with frequencies higher than 20 % (H1 and H2), three haplotypes (H3, H4 and H5) found at low frequencies but in multiple populations, and nine unique haplotypes (HU1 HU9) found in single individual lark sparrows (Fig. 1; Table 1). One of the common haplotypes (H1) was found in all sampled breeding populations (n = 134) while the other high frequency haplotype (H2) was mostly present in the Texas population (n = 63) but also found in one individual in Nebraska. H3, H4 and H5 were all low frequency haplotypes found in TX, NE, IL and OH. Unique Haemoproteus haplotypes were detected in the Ohio, California, and Texas populations. The presence of Plasmodium was rare in all populations (Table 1). A total of seven Plasmodium haplotypes were detected (GenBank accession numbers KJ735813 KJ735819), with three being found only in single individual lark sparrows. Furthermore, only two Plasmodium haplotypes (P2 and P3) were found
242 Genetica (2014) 142:235 249 Fig. 3 Haplotype networks for Haemoproteus and Plasmodium avian malaria parasites, with confidence limits for creating connections set at 95 %. The area of the circles is proportional to the haplotype frequency. Populations where haplotypes were found and haplotype identification are indicated. Black dots represent intermediate haplotypes not found in our study in more than one population. There were also differences in the haplotypes found in the different age classes of lark sparrows. In Texas, four haplotypes (H1, H2, H4 and H5) were found in the HY birds. In Ohio, Haemoproteus was mostly restricted to the AHY birds while Plasmodium was mainly restricted to the HY individuals. All malaria haplotypes were subjected to BLAST searches to determine if they were restricted to the lark sparrow or found in other bird hosts (Table 2). The two most prevalent Haemoproteus haplotypes (H1 and H2) did not have any complete matches in the GenBank database (out of 1,525 Haemoproteus cytochrome b sequences reported to GenBank as of April 2014). Haplotypes HU2 through HU9 also failed to have any exact matches. Haplotypes H3 and HU1 were the only Haemoproteus haplotypes with exact matches to sequences detected in Sialia mexicana, Sialia sialis, and Dendroica pensylvanica. While most of the Haemoproteus lineages did not have any exact matches, the opposite was true for Plasmodium haplotypes. Only two haplotypes (PU1 and PU2) did not have any exact matches, but the remaining Plasmodium haplotypes (P1 P4 and PU3) had matches with sequences from a variety of host species (out of 4,954 Plasmodium cytochrome b sequences reported to GenBank as of April 2014; see Table 2). Genetic variation was found to be higher in the Plasmodium (p-distance = 0.047; SE = 0.008) compared to the Haemoproteus sequences (p-distance = 0.014; SE = 0.004) and, as expected, was highest between the two genera of malaria parasites (p-distance = 0.103; SE = 0.016). When considering regional differences, average levels of diversity
Genetica (2014) 142:235 249 243 Table 3 Genetic diversity estimates for malaria parasites (Haemoproteus and Plasmodium) by breeding population of lark sparrow Populations are listed by state abbreviation codes. Sample sizes refer to the number of sequences of each specific parasite obtained from each population. Estimates include mean values ± standard deviations Population Sample size Number of haplotypes Mean number pairwise differences Haplotype diversity Nucleotide diversity Haemoproteus CA 22 5 0.80 ± 0.60 0.34 ± 0.13 0.003 ± 0.002 NE 19 3 5.32 ± 2.68 0.37 ± 0.13 0.018 ± 0.010 TX 111 8 4.94 ± 2.42 0.58 ± 0.03 0.016 ± 0.009 IL 17 3 0.24 ± 0.29 0.23 ± 0.13 0.001 ± 0.001 OH 54 6 0.85 ± 0.61 0.18 ± 0.07 0.003 ± 0.002 Overall 223 14 4.22 ± 2.10 0.56 ± 0.03 0.014 ± 0.008 Plasmodium CA 3 1 0 0 0 NE 1 1 0 0 0 TX 2 2 6.00 ± 4.58 1.00 ± 0.50 0.020 ± 0.010 IL 1 1 0 0 0 OH 15 4 11.28 ± 5.42 0.74 ± 0.06 0.037 ± 0.006 Overall 22 7 13.51 ± 6.31 0.84 ± 0.04 0.045 ± 0.004 within regions (p-distance = 0.027; SE = 0.004) were not significantly different than that of the entire population (pdistance = 0.029; SE = 0.005), which resulted in relatively low levels of inter-region diversity (p-distance = 0.002; SE = 0.002). The connection of haplotypes into mutational networks using a 95 % CI resulted in the characterization of two Haemoproteus network groups while the Plasmodium haplotypes were divided into four independent networks separated by more than 10 mutational differences (Fig. 3). The main Haemoproteus network (which included all but two haplotypes) revealed a star shape with the most common lineage (H1) in the center. The network is fairly tight with no more than three intermediate haplotypes being inferred between observed haplotypes. Only two Haemoproteus haplotypes clustered independently from the starshaped network. In contrast, the Plasmodium haplotypes were grouped into two linear networks with a larger number of mutational steps connecting haplotypes and two distinct haplotypes separated by more than 10 mutational differences from any other haplotype. Given the limited number of Plasmodium haplotypes detected across populations, the analysis of malaria genetic structure was restricted to the Haemoproteus genus. Overall, host populations had significant differences in their levels of avian malaria genetic diversity as measured by haplotype and nucleotide diversities (Table 3); however, some of these differences may be due to differences in sample sizes. AMOVAs based on Haemoproteus haplotype distances and frequencies both indicated significant levels of genetic diversity among populations, with 15 18 % of the total variation being explained by differences among populations within regions (Table 4). Although 18 % of the observed variation was assigned to Table 4 Hierarchical analysis of molecular variance (AMOVA) based on Haemoproteus parasite haplotype sequences Source of variation Pairwise difference Haplotype frequency Variance component Percent variation Variance component Percent variation Among regions 0.439 18.07 0.058 17.93 Among populations 0.375 15.41* 0.059 18.26* within regions Among individuals within populations 1.616 66.52* 0.207 63.82* AMOVAs were performed independently using genetic distances based on pairwise differences and haplotype frequencies * p \ 0.001 differences among regions, this value was not statistically significant. Pair-wise estimates of population differentiation revealed that the Texas population was significantly different from each of the other four populations studied. The phylogenies for the lark sparrow malaria parasites produced by Bayesian methods revealed evolutionary relationships among parasite lineages (Figs. 4, 5). As expected, the general phylogeny including malaria parasites from multiple avian species showed Haemoproteus and Plasmodium sequences forming reciprocally monophyletic groups (Fig. 4). The Plasmodium haplotypes found in the lark sparrow did not cluster as an independent monophyletic group. Instead sequences were dispersed throughout the tree grouping independently with parasite lineages from other avian species. In contrast, Haemoproteus haplotypes from the lark sparrow tended to cluster into a few independent monophyletic lineages within the general phylogeny (Fig. 4).
244 Genetica (2014) 142:235 249 Fig. 4 Evolutionary relationships of avian malaria parasites from the lark sparrow and other bird species inferred using Bayesian analysis. Numbers at nodes represent Bayesian posterior probabilities. Gray boxes indicate haplotypes found in the lark sparrow. Haplotypes from other species are listed by their GenBank accession numbers The phylogeny of avian malaria parasites from the lark sparrow allowed for assessment of the phylogeographic distribution of malaria haplotypes across populations (Fig. 5). As expected, Plasmodium and Haemoproteus sequences formed clearly distinct monophyletic groups. At least three major lineages (i.e., major monophyletic groups with posterior probabilities [0.79) were identified within each of the Plasmodium and Haemoproteus genera. Only one malaria haplotype from Haemoproteus (H1) was detected in all populations studied. Although CA, TX and OH showed all population-specific haplotypes, no major phylogeographic patterns associating major malaria lineages to specific host populations were evident (Fig. 5). Furthermore, none of the populations had representative haplotypes from all major lineages. TX and OH, however, were the most variable populations, with the greatest number of haplotype representatives. IL and NE populations revealed the least number of haplotypes (Figs. 3, 5). Discussion Results from this study showed that levels of avian malaria prevalence were consistently high across breeding populations of the lark sparrow. The overall prevalence of avian malaria in adult lark sparrows was over 80 % (n = 254). When compared to published infection rates for other species, the prevalence across the lark sparrow s range is notably high. For example, in a study of the black-throated blue warbler (Dendroica caerulescens) across its North
Genetica (2014) 142:235 249 245 Fig. 5 Evolutionary relationships of avian malaria parasites from the lark sparrow inferred using Bayesian analysis. Numbers above each node represent Bayesian posterior probabilities. Checkmarks indicate the specific population(s) in which each haplotype was found American breeding range, overall malaria prevalence was 22 % with infection rates varying from 14 to 43 % by region (Fallon et al. 2006). Similarly, the American redstart (Setophaga ruticilla) was shown to have overall malaria prevalence of 33 % on its breeding grounds with prevalence by region varying from 22 to 54 % (Durrant et al. 2008). One breeding region of the common yellowthroat (Geothlypis trichas) had malaria prevalence of 78 %, but overall prevalence was 53 % and prevalence in the other four breeding regions varied from 20 to 64 % (Pagenkopp et al. 2008). There were no significant differences in prevalence in adult lark sparrows by population or by region. Temporal differences in populations sampled over three sampling years were also not apparent. There was, however, a significant difference in malaria prevalence between adult males and females, with females having higher infection rates. This trend was consistent across regions, though statistical significance was detected only in the central populations. Although other studies have reported sex differences in malaria prevalence and infection effects (Korpimäki et al. 1993; Ortego et al. 2008; Lachish et al. 2011a, b), results do not show consistent
246 Genetica (2014) 142:235 249 trends. The general pattern observed in this study may be related to the fact that in the lark sparrow, females devote more time than males to incubation, potentially leaving them more exposed to vectors and infections. Overall, significant differences in malaria prevalence were found between age classes of host birds, with adults having higher rates of infection than HY juveniles. Population-specific significant differences in prevalence by age class were found in the California and Ohio populations. Increased malaria infections between first year and older birds have also been reported in blue tits (Wood et al. 2007; Lachish et al. 2011a, b). Several factors can potentially contribute to these differences. For example, differences in the rate of infection between adult and juvenile birds could be related to exposure time. The juveniles sampled were at most a few weeks from fledging, so there may not have been much opportunity for exposure to suitable vectors or enough time for the infection to reach the peripheral blood. The location where infections are occurring could also be a major factor (Szöll}osi et al. 2011; Loiseau et al. 2012). The fact that juveniles were infected indicates that some infections were effectively occurring on the breeding grounds, but it is thus far unknown what proportion of infections occurs on breeding versus wintering grounds. In the Ohio population it appears that mainly Plasmodium parasites are being transmitted on the breeding grounds, as only one of 14 HY individuals was infected with Haemoproteus parasites. This may imply that the common Haemoproteus haplotype that is found across all populations is being transmitted on the lark sparrows common wintering grounds or that infected birds are effectively migrating from other regions/populations. Recent studies have shown that the distribution and prevalence of avian malaria parasites varies considerably depending on both host and parasite characteristics (e.g., Szöll}osi et al. 2011). In particular, there have been reported differences on the degree of host specificity between Plasmodium and Haemoproteus parasites (Atkinson and van Riper 1991; Pérez-Tris et al. 2005; Hellgren et al. 2009). Furthermore, Pérez-Tris et al. (2005) found a relationship between dispersal and local transmission rates of parasites, with year-round transmission facilitating spread of parasites throughout migratory avian hosts. This is consistent with a study by Hellgren et al. (2009), which showed that malaria parasites with broader host range compatibility were also the most prevalent within single host species. The genetic characterization of avian malaria parasites revealed that the majority of infections occurring in the lark sparrow were found to be with Haemoproteus parasites. The most common haplotype (H1) was found in all five of the breeding populations, consistent with the idea that parasite lineages with wider distribution had higher local prevalence (Szöll}osi et al. 2011). However, the second most prevalent haplotype detected in the lark sparrow (H2) was found predominantly in Texas and, in fact, made up the majority of Texas infections (55 %), suggesting that other factors may be driving parasite prevalence in this population. Of particular relevance is the fact that some of the host populations in Texas tend to be non-migratory and therefore potentially exposed to a distinct parasite population all year round. A small number of samples revealed double infections, mostly with two distinct Haemoproteus haplotypes; however, two individuals showed mixed infections with Haemoproteus and Plasmodium. Although studies from wild populations usually report low incidence of mixed infections, this may be due to technical issues associated with the detection of distinct malaria lineages by PCR (Bensch et al. 2007). Most Haemoproteus haplotypes reported here appear to be restricted to the lark sparrow, as inferred by BLAST searches, which revealed no exact matches in the GenBank database for the most common haplotypes detected in this study. In fact, only two (HU1 and H3) of the 14 Haemoproteus haplotypes found in the lark sparrow have also been found in other avian species (Table 2). Furthermore, many of the Haemoproteus haplotypes detected in the lark sparrow grouped together in a general phylogeny including malaria parasites from multiple avian species (Fig. 4). This suggests that at least some Haemoproteus lineages may be host-specific. In contrast, Plasmodium infections were rare and found mainly in Ohio, with the most prevalent haplotype found in six individuals. When Plasmodium sequences were subjected to BLAST searches, five of the seven haplotypes had exact matches with sequences from other host species. In fact, the general phylogeny of avian malaria parasites revealed that Plasmodium haplotypes found in the lark sparrow were dispersed throughout the phylogenetic tree, clustering with malaria parasites/lineages detected in other avian hosts. These results suggest that, compared to Haemoproteus, Plasmodium lineages may represent generalist morphospecies, able to infect a wider range of host bird species. This is consistent with previous studies, which have suggested that Haemoproteus infections are more host-specific than Plasmodium (Atkinson and van Riper 1991; Beadell et al. 2004; Hellgren et al. 2009). Differences in host specificity are likely related to differences in resistance mechanisms and to the specific vectors involved in transmitting Plasmodium versus Haemoproteus parasites. Despite smaller sample size and distribution, Plasmodium haplotypes showed more genetic variation than Haemoproteus haplotypes. Overall, Plasmodium had higher levels of diversity in terms of mean number of pairwise differences, haplotype diversity, and nucleotide diversity (Table 3). This is not surprising since
Genetica (2014) 142:235 249 247 Plasmodium haplotypes may represent highly divergent species, as shown by their independent clustering with parasite lineages identified in other host species (Fig. 4). In contrast, several Haemoproteus haplotypes grouped as independent monophyletic groups (e.g., H4 HU6; H2 HU3 HU7). These parasite lineages may have originated through speciation events, either following subdivision of parasite populations in discrete habitats or sympatrically through colonization of different host species (Pérez-Tris et al. 2007). Differences in the levels of diversity are also consistent with the information from the BLAST searches, indicating that Haemoproteus parasites are more hostspecific (and thus show less variation) while Plasmodium parasites tend to be more generalists. Genetic diversity of Haemoproteus parasites seems to be structured by population, as evidenced by the results of the AMOVA, which showed that 15 18 % of the total genetic variation can be explained by differences among host populations. However, there were not significant differences in Haemoproteus genetic variation among lark sparrows breeding regions. This result is likely related to the high variation in haplotype frequencies detected among host populations and the distribution of population-specific haplotypes (Fig. 5), which may either prevent or mask potential regional effects (Fallon et al. 2006). On the other hand, the detection of population-specific haplotypes may be related to the general evolutionary tendency of generalist parasite lineages to become more specialists (Loiseau et al. 2012). The significant genetic structuring of Haemoproteus parasites across populations may therefore be driven by the specific dynamics of the host population and/ or the local vector distribution and abundance in each geographic area. Significant differentiation in both morphological and genetic characteristics of lark sparrow subspecies associated with the western, central, and eastern breeding regions (Ross and Bouzat 2014), suggests that geographic isolation may play an important role in promoting demographic independence between parasite host populations. The lack of concurrent differentiation of malaria parasites by region indicates that the population dynamics of the parasites may be decoupled from the regional population structure of lark sparrows. The demographic dynamics of avian malaria may instead be driven at smaller spatial scales (e.g., at the population level of avian hosts). This is evidenced by the relatively large number of population-specific haplotypes of both Haemoproteus and Plasmodium parasites. The presence of a common Haemoproteus haplotype (H1) distributed across all sampled breeding populations suggests, however, a distinct demographic history from that revealed by the Plasmodium haplotype distribution, which showed no common haplotypes across host populations. These differences are likely driven by the specificity of Haemoproteus lineages associated with the lark sparrow. This is also reflected by the star shape of the main Haemoproteus haplotype network, characteristic of rapid demographic expansions, in contrast to the linear networks separating haplotypes with multiple mutational steps detected in Plasmodium. This study shows that avian malaria prevalence across the breeding range of the lark sparrow is considerably high compared to previously reported rates on other avian species (with up to 81 % of breeding individuals testing positive for avian malaria). The detection of malaria parasites in both adult and juveniles of the lark sparrow and the differences in their prevalence revealed that malaria infections are effectively occurring on the lark sparrow breeding grounds, but suggest that the dynamics of infection vary among populations and, potentially, between breeding and overwintering grounds. Furthermore, levels of genetic diversity and genetic structure of malaria parasites with respect to the host populations revealed distinct patterns for Haemoproteus and Plasmodium, likely as a result of their distinct life histories and transmission vectors (Atkinson and van Riper 1991; Bensch et al. 2000; Beadell et al. 2004; Wood et al. 2007; Lachish et al. 2011a, b). In contrast to the regional patterns of genetic differentiation previously detected in the lark sparrow (Ross and Bouzat 2014), Haemoproteus showed high levels of population-specific variation with no regional differentiation, which suggest that the population dynamics of the parasites may be driven by evolutionary processes operating at small spatial scales, at the level of host populations. These results highlight the importance of considering the potential effects of host and parasites population dynamics on the distribution and prevalence of avian malaria, an essential issue in understanding the potential effects of wildlife diseases on avian host species. Acknowledgments We thank Jeremy D. Ross for coordinating collection of blood samples of lark sparrows, V. P. Bingman, M. Herman, T. Herman, E. Keller, J. Noland, and E. A. Ross for field assistance, and Elijah Bodey and Hanna Scheppler for helping in the molecular sexing of birds. We would also like to thank the Editor and three anonymous reviewers for their suggestions on previous versions of the manuscript. We thank the following institutions/individuals for allowing field access to Lark Sparrow populations: Oak Openings Preserve, Metroparks of the Toledo Area; Kitty Todd Nature Preserve, The Nature Conservancy; Lost Mound Unit of Upper Mississippi National Wildlife Refuge, US Fish and Wildlife Service; Cedar Point Biological Station, University of Nebraska at Lincoln; Daniel Brown and Skipper Duncan, Tom Green County, Texas; Hopland Research and Extension Center, University of California at Davis; and Gray Davis/Dye Creek Preserve, The Nature Conservancy; R. C. Dawkins, Angelo State University; D. G. Wenny, Illinois Natural History Survey; and M. C. Shieldcastle, Ohio Department of Natural Resources. This project was funded through State Wildlife Grants from the Ohio Department of Natural Resources Division of Wildlife, in-kind contributions from The Nature Conservancy, and the Department of Biological Sciences at Bowling Green State University.