Introduction

The molecular phylogeny of the coccidia has been the subject of numerous studies and opinions (Cavalier-Smith 1983, 1993; Tenter et al. 2002). However, it has become necessary, in the light of recognized shortcomings of previous studies, to explore more genes and genomes in order to have a better understanding of the relationships within the phylum Apicomplexa, particularly with respect to the coccidia (Tenter et al. 2002; Morrison et al. 2004; Morrison 2008, 2009).

The difficulty of establishing a widely accepted molecular phylogenetic hypothesis for the phylum Apicomplexa is compounded by the fact that classical as well as some current phylogenetic studies have been based on phenotypic characters or limited molecular data. The molecular data for inferring phylogeny in Apicomplexa have been limited in most cases to the nuclear genome and frequently only 18S rDNA sequences (Carreno and Barta 1999; Morrison et al. 2004; Morrison 2008). The difficulty of establishing positional homology among18S rDNA sequences across divergent taxa has complicated the use of these sequences for resolving phylogenetic relationships within the Apicomplexa and other phyla (Barta et al. 1997, 2001; Carranza et al. 1996; Li et al. 1997). Other nuclear genes that have been used to investigate the evolutionary history of the Apicomplexa include 28S rDNA, ribosomal ITS regions, and adenylosuccinate lyase (Mugridge et al. 1999; Kedzierski et al. 2002; Martinsen et al. 2008; Samarasinghe et al. 2008).

Most members of the Apicomplexa possess functional nuclear, mitochondrial, and plastid genomes; however, gregarines apparently lack a plastid genome, and Cryptosporidium spp. lack both plastid and mitochondrial genomes (Wilson and Williamson 1997). Tenter et al. (2002) suggested that only analyses using multiple genes, preferably including both nuclear and organellar genes, could generate a molecular phylogeny for the eimeriid coccidia that is likely to represent a reasonable organismal evolutionary hypothesis. The lack of comparable genes and genomes among some members of the Apicomplexa and variation in the evolutionary rates of different genes within genomes can be problematic. Molecular data generated from multiple genes and genomes should be subjected to parametric evolutionary models and robust phylogenetic analysis in order to establish homology in support of the interpretation of tree topologies as evolutionary histories, especially in the Apicomplexa (see Morrison (2008)). The use of multiple genes and genomes for molecular phylogenetic and epidemiological studies in the Apicomplexa has been examined by some workers (Rathore et al. 2001; Perkins et al. 2007; Schwarz et al. 2009; Hikosaka et al. 2010). Waller and McFadden (2005) examined the origin and function of the plastid in some members of the Apicomplexa. Phylogenetic studies involving members of the Apicomplexa have been carried out using plastid genes such as ORF472 (open reading frame), CLPs (caseinolytic proteases), elongation factor Tu (TufA), and ribosomal polymerases in the eimeriids and haemosporinids ( Lang-Unash et al. 1998; Blanchard and Hicks 1999; Cai et al. 2003; Saxena et al. 2007). Zhao et al. (2001) have suggested that the plastids might be a useful source of markers for the evolutionary delineation of apicomplexan taxa that possess this genome.

The commonest mitochondrial genes used for phylogenetic studies are cytochrome oxidases because of their universal occurrence in organisms that utilize oxidative phosphorylation as an energy source (Hebert et al. 2003, 2004). There have been few phylogenetic studies involving coccidia using mitochondrial genes (Schwarz et al. 2009; Ogedengbe et al. 2011), but mitochondrial gene sequences have been used extensively and successfully with haemosporinids and piroplasms (e.g., Putignani et al. 2004; Omori et al. 2007; Perkins 2008) for both molecular systematics and epidemiology. Perkins et al. (2007) suggested that genes residing on each of the three genomes could be useful for constructing phylogenetic trees and can serve as a means of comparing evolutionary rates and patterns of sequence evolution through phylogenetic inference. The use of total evidence (all available datasets) in the study of phylogeny is thought to be advantageous (Eernisse and Kluge 1993; de Queiroz et al. 1995). This study is a first step in evaluating the use of both rRNA and protein-coding genes from the three apicomplexan genomes (plastid, mitochondrial, and nuclear) in phylogenetic analyses of a number of coccidian parasites (Apicomplexa, Coccidia).

Materials and methods

Sources of parasites and parasite DNA

Oocysts of coccidia were obtained from a variety of sources. Laboratory strains of Eimeria spp. from chickens (either single sporocyst- or single oocyst-derived lines maintained in our laboratory) were propagated in specific parasite-free chickens in the OMAFRA animal isolation facility (University of Guelph, Guelph, ON, Canada); chicks were provided with feed and water ad libitum. All experimental procedures were reviewed and approved by the University of Guelph’s Animal Care Committee and complied with the Canadian Council on Animal Care’s Guide to the Care and Use of Experimental Animals (second edition). Sporulated oocysts of laboratory strains of the murine parasites Eimeria falciformis and Eimeria papillata were kindly provided by Dr. Bill Chobotar, Andrews University, Berrien Springs, MI, USA. Sporulated oocysts of the marsupial parasite, Eimeria trichosuri, were kindly provided by Dr. Michelle Power, Macquarie University, Sydney, Australia. Sporulated oocysts of Cystoisospora suis from swine, Cystoisospora felis from domestic cats, and Eimeria zuernii from cattle were obtained from clinical fecal specimens submitted for diagnosis to the Animal Health Laboratory (University of Guelph, Guelph ON, Canada). Sporulated oocysts were concentrated from fecal debris using standard salt flotation methods (Reid and Long 1979) and finally stored in 2.5 % potassium dichromate (w/v aqueous) at 4 °C prior to DNA extraction.

Purified DNA from a laboratory culture of Toxoplasma gondii strain ME49 was kindly provided by Dr. J. P. Dubey, USDA, Beltsville, MD, USA.

DNA extraction

Sporozoites were excysted from sporulated oocysts by first grinding them to release sporocysts, and sporozoites were released by incubation at 42 °C in excystation fluid containing either bile-trypsin for oocysts of Eimeria maxima or taurocholic acid-trypsin for other types of oocysts. DNA was extracted from cleaned sporozoites by using DNAzol (GIBCO- Life Technologies, USA) according to the manufacturer’s protocol. DNA aliquots (140 ng/ul) were prepared and frozen, while working DNA was diluted to about 30 ng/ul. Polymerase chain reactions (PCRs) were performed using 60–90 ng template DNA.

PCR

PCR was used to amplify specific plastid and mitochondrial genes using the primers shown in Table 1. One unit of Platinum®Taq DNA Polymerase, 0.5 mM dNTP’s, 1×PCR buffer, and 2.5 mM MgCl2 (Invitrogen, USA) was used to amplify ∼90 ng DNA template in each 50 μl reaction. The reaction conditions were an initial denaturation at 95 °C for 6 min followed by 35 cycles of 95 °C for 20 s, 55 °C for 30 s, and 72 °C for 1 min, with a final extension at 72 °C for 10 min. Both negative and positive template control reactions were included with each PCR run. PCR products were electrophoresed on a 1.5 % agarose submarine gel in 1×Tris-acetate-EDTA buffer at 120 V for 45 min. The resulting gel was stained with ethidium bromide, and the size of the products was estimated by comparison with a 100-bp DNA size standard (Invitrogen).

Table 1 PCR primers used to amplify partial plastid and mitochondrial genes
Full size table

Phylogenetic analysis

Sequencing and sequence alignments

PCR products were either purified directly or from agarose gels using the Roche High Pure PCR Product Purification Kit® (Roche Applied Science, Germany) according to the manufacturer’s instructions. PCR cycle sequencing using the forward or reverse amplification primers was followed by detection on an ABI Prism 3730 or 3100 DNA sequencer (Molecular Biology Unit of the Laboratory Services Division, University of Guelph). High-quality contigs and final consensus sequences were generated using Geneious Bioinformatics software package (Version 6.1.8) (http://www.geneious.com, Kearse et al. 2012). Each cytochrome c oxidase subunit I sequence was approximately 500–810 base pairs in length depending on the primer set used to amplify this gene (Table 1). The 500-bp PCR products covered the 5′ end of the 800-bp PCR product, so all sequences were from the same region of the COI gene. The protein translation of the sequences presented open reading frames without stop codons. New and existing published COI sequences were aligned using Clustal-X (Larkin et al. 2007) and MAFFT (Katoh et al. 2009) implemented from within the Geneious (Ver. 6.1.8) Bioinformatics software package.

A secondary-structure-based nu 18S rDNA sequence alignment was generated within Geneious using the secondary structure map of T. gondii (see Gagnon et al. 1996) to assign structural annotations. The entire alignment was then edited manually by staggering non-homologous alignments to minimize obvious misalignments according to the method described by Barta (1997). Phylogenetic analysis of the nu 18S rDNA was performed using a wide variety of apicomplexan taxa; corresponding GenBank accession numbers are found on all phylogenetic trees.

A trio of new COI sequences was added to 115 published sequences for the analyses (Cystoisospora felis OVC_2008, JN227478; Eimeria praecox Guelph_2010, JQ659301; Eimeria trichosuri, JN192136). GenBank accession numbers for all other sequences used in the COI and mixed genome analyses are documented on the resulting phylogenetic trees.

A total of 37 new sequences from 15 species from up to four different plastid loci were generated in the present study: T. gondii ME49 (JN181049); Cystoisospora felis OVC_2008 (JN181039), Cystoisospora suis Guelph 2008 (JN181040), E. acervulina (JN181038, JN181028, JN181043, JN181059), E. brunetti (JN181036, JN181050, JN181053), E. falciformis (JN181037, JN181032, JN181044, JN181052), E. mitis USDA 50 (JN181030, JN181042, JN181060), E. maxima (JN181034, JN181027, JN181047), E. necatrix (JN181035, JN181029, JN181045, JN181058), E. papillata (JN181046, JN181056), E. praecox (JQ659302, JQ659303), E. tenella (JN181033, JN181031, JN181041, JN181054), E. trichosuri (JN181051, JN181057), E. vermiformis (JN181048, JN181061), and E. zuernii (JN181055). The GenBank accession numbers of all generated and previously published sequences are found on the resulting phylogenetic trees. The plastid sequence data associated with a particular parasite species consisted of up to four concatenated sequences of partial 16S rDNA and 23S rDNA sequences and ribosomal polymerase (rpoB and rpoB1) sequences. All 37 new plastid sequences obtained were confirmed to be similar to known parasites of the same genus using nucleotide BLAST (Altschul et al. 1990) searches in GenBank. Sequences were next assembled and aligned using Geneious (Kearse et al. 2012). Alignment was initially performed using MAFFT and Clustal W (Katoh et al. 2009; Larkin et al. 2007). Protein-coding sequences were translated to confirm that open reading frames existed without stop codons.

Sequence datasets used for phylogenetic analyses

A total of four datasets were generated for the analyses. The first three datasets consisted of all publicly available and newly generated sequences for Eimeria species and the outgroup taxa: dataset 1—all available 18S rDNA sequences, dataset 2—all mt COI sequences, and dataset 3—concatenation of up to four plastid genes (SSU, LSU, rpoB, and rpoB1). The final total evidence dataset from all genomes was constructed as follows: For each parasite species, a single strict consensus sequence was generated of its nu 18S rDNA, its mt COI, and a concatenation of up to four of its plastid genes (described above). The resulting three consensus sequences for each genome for each parasite species (including ambiguity codes at variable positions among multiple sequences from a single genetic locus in a single parasite species) were concatenated into a single combined evidence dataset.

Analyses were performed using maximum likelihood and maximum parsimony in PAUP (Swofford 2003) and Bayesian analysis (Huelsenbeck and Ronquist 2001) with parameters estimated from model tests. Analyses of structure-based alignment of nu 18S rDNA sequences were attempted using regions involved in pair bonding (helices), regions without pair bonding (non-helices), or all data. Five million generations were run for the first three datasets (all 18S rDNA [three iterations based on structure]; COI and plastid sequences), whereas1.5 million generations were sufficient for convergence in the remaining datasets. In all datasets analyzed, a comprehensive model test was performed to determine the best nucleotide substitution rate model using MrModeltest (Nylander 2004). In all tests, the hierarchical likelihood ratio test was employed, and the best model was selected using the Akaike information criterion.

Results

A total of 42 new sequences from various genetic loci were generated in support of these analyses; all sequences have been submitted to GenBank as outlined above.

18S rDNA sequence analysis

The best-fit model for the global 18S rDNA dataset was the GTR+I+G. Using subsets of the aligned nu 18S rDNA sequence data (i.e., helices-only characters excluded or non-helices-only characters excluded) did not improve phylogenetic resolution of the resulting evolutionary hypotheses compared with analyses based on the complete18S rDNA datasets (data not shown). For this reason, the complete nu 18S rDNA sequence dataset was used in all further analyses incorporating the GTR+I+G substitution model.

The ML, MP, and Bayesian analysis trees were topologically similar, but only the trees generated from Bayesian analysis and their posterior probabilities are shown. The MP tree of the complete 18S rDNA sequence dataset had 223 taxa with a total of 2023 characters weighted equally; 450 of these characters were parsimony informative. The tree length was 2430 with a consistency index of 0.5. The Bayesian analysis generated a consensus tree (see Fig. 1) in which members of the Toxoplasmatinae (species of Cystoisospora, Neospora, Hammondia, and Toxoplasma) formed a well-supported monophyletic group that had a sister group relationship with coccidia of poikilothermic vertebrates (e.g., Hyaloklossia and Goussia spp.) and all other coccidia. Goussia metchnikovi and Eimeria tropidura found in fish and lizards respectively formed a well-supported clade. Members of the Toxoplasmatinae and all of the early branching coccidia (Hyaloklossia and Goussia species plus Eimeria tropidura) illustrated in Fig. 1 possess valvular sutures on their sporocysts and do not have Stieda bodies. Eimeria arnyi from a snake and E. ranae from frogs were early branching to a trichotomy comprised of (1) Eimeria species of marsupials (E. trichosuri), (2) Eimeria spp. of cranes (E. gruis and E. reichenowi), and (3) the remaining coccidia. The genus Eimeria was observed to be polyphyletic with at least four independent lineages of Eimeria species being supported by the sequence data. Caryospora and Lankesterella species formed a monophyletic clade with 0.83 posterior probability support. Avian Isospora and Atoxoplasma species formed a well-supported monophyletic clade (1.00 posterior probability support—not shown in Fig. 1), although the monophyly of neither of these genera was supported. Eimeria species frequently formed well-supported clades of parasites that parasitized the same or closely related definitive hosts such as Eimeria spp. infecting swine (1.00 posterior probability support), rabbits (1.00 posterior probability support), or galliform birds (1.00 posterior probability support) (see Fig. 1).

Fig. 1
figure 1figure 1figure 1

Consensus Bayesian phylogenetic tree generated from 223 18S rDNA sequences (GTR+I+G) from 94 taxa with posterior probabilities of node support indicated. The same tree was obtained with MP analysis with a tree length of 2430 and a consistency index of 0.5. Inset: Overview of the complete tree with the 3 enlarged regions indicated

Full size image

Eimeria alabamensis from cattle was the sister taxon to all Eimeria found in ruminants (mainly ovine) and lagomorphs, and both were sister to the species found in rodents or pigs. Cyclospora species formed a well-supported monophyletic clade that was a sister group to a large clade of Eimeria species that infect piciform and galliform birds.

Cytochrome c oxidase subunit I analysis

The GTR+G+I-based codon model using a mitochondrial translation table (code=metmt) was applied to the COI sequences. Figure 2 shows the tree generated from the Bayesian analysis. Similar trees were generated using both MP and ML analyses performed using 93 sequences made up of 833 characters, of which 298 were constant, 115 were variable, and 420 were parsimony informative; for the MP analysis, tree length was 1339, and CI was 0.61. The taxonomic outgroup was comprised of members of the Toxoplasmatinae (Cystoisospora, Neospora, and Toxoplasma spp.) and formed the sister group to the Eimeria species. In cases where there were multiple sequences from an individual Eimeria sp., all sequences formed a well-supported (PP=1.00) monophyletic clade with markedly lower intraspecific genetic distances compared to the 18S rDNA tree described above. Similarly, E. mitis, E. mivati, and E. cf mivati sequences formed a monophyletic “species” grouping. Eimeria species from chickens did not form a monophyletic clade; Eimeria sp. sequences from other birds (Chukar partridge, turkeys, and pheasants) formed a weakly supported monophyletic clade with E. tenella and E. necatrix that excluded the other Eimeria species of chickens. Although the relationships among closely related taxa were well resolved, the relationships among the species clusters were poorly resolved as evidenced by the large basal polychotomy.

Fig. 2
figure 2

Consensus Bayesian tree generated from 101 cytochrome oxidase c subunit I sequences (GTR+G+I; codon model; code=metmt) with posterior probabilities of node support indicated. Posterior probabilities for nodes within monophyletic species-level clades are not shown. The same tree topography was obtained with MP analysis with a tree length of 1339 and a consistency index of 0.61

Full size image

Analysis of concatenated plastid gene sequences

The best-fit model for both partitions of the plastid gene sequence dataset (ribosomal RNA, 1014 bp; RPOB sequences, 1496 bp) was the GTR+G+I substitution model; the rRNA sequence partition used a nucleotide model, whereas the RPOB partition used a codon model (code=metmt). Figure 3 shows the tree generated from Bayesian analysis from 23 sequences. Both MP and ML gave the same tree topologies. The MP analysis performed on the 23 taxa included 2473 characters, 1484 of which were constant, with a tree length of 1462 and CI of 0.80, and where 757 were parsimony informative. The sequences representing the four plastid genes, where available, were concatenated for each taxon. With members of the Toxoplasmatinae as outgroup, the branching order of several of the Eimeria species used in the analysis was unexpected. Several Eimeria spp. from bats (E. arizonensis, E. antrozoi, and E. rioarribaensis) formed a sister group to a number of Eimeria spp. found in mammals (rodents and herbivores) as well as, unexpectedly, E. maxima. Eimeria brunetti was sister taxon to a well-supported monophyletic clade consisting of E. necatrix and E. tenella.

Fig. 3
figure 3

Consensus Bayesian tree generated from 23 concatenated available plastid sequences (LSU, SSU, rpoB1, and rpoB) with posterior probabilities of node support indicated. Bayesian analysis used GTR+I+G with a nucleotide model for rDNA sequences and GTR+G+I with codon model (code=metmt) for rpoB sequences. The same tree topography was obtained with MP analysis with a tree length of 1,014 and a consistency index of 0.73

Full size image

Multiple gene and genome consensus tree

Figure 4 shows the tree generated by Bayesian analysis of concatenated consensus sequences from all taxa that had at least one representative sequence in all genomes. From the 12 taxa that were used, the MP tree length was 1030 with 2614 characters, of which 2002 are constant and 476 were parsimony informative with a CI of 0.72. The Eimeria species that infect birds formed a monophyletic clade that was sister to both E. trichosuri of marsupials and E. falciformis of rodents with which they formed a trichotomy.

Fig. 4
figure 4

Consensus Bayesian “total evidence” tree generated (GTR+I+G) from 12 taxa of concatenated strict consensus sequences from plastid, mitochondrial, and nuclear gene sequences with posterior probabilities of node support indicated. The same tree topography was obtained with MP analysis with a tree length of 1859 and a consistency index of 0.76

Full size image

Discussion

The use of single or multiple genes as markers in the molecular phylogeny of organisms is an important aspect of molecular systematics (Avise 1994). In this study, we generated novel sequences from the mitochondrial and plastid genomes of some members of the Apicomplexa to make a preliminary evaluation of the usefulness of a multigene and multigenome approach to molecular phylogeny of the eimeriorinid apicomplexan parasites.

The use of multiple genomes and genes as genetic markers to study the molecular phylogeny of parasites has been the subject of many studies (Escalante and Ayala 1994; Siddall et al. 1997; Rathore et al. 2001; Kedzierski et al. 2002; Leander et al. 2003; Hagner et al. 2007; Lau et al. 2009; Bhoora et al. 2009; Hikosaka et al. 2010; Outlaw and Ricklefs 2010). An advantage of combining data from different genomes or different genes is that genetic loci can be selected so that they differ in their rates of evolutionary change (Yang 1996).

In this study, phylogenetic analyses using single gene sequences from nuclear (18S rDNA), mitochondrial (COI), and concatenated plastid (SSU, LSU, RpoB and RpoB1) genes reflected some of the difficulties encountered when using genetic markers from single genomes, such as difficulties in establishing positional homologies, particularly with the 18S rDNA sequences that are prone to large variations in sequence length (see Barta 2001). Analysis using a concatenation of consensus sequences from all available genes in a total evidence approach appeared able to resolve clades of Eimeria species in birds, marsupials, and rodents more effectively than the use of sequences from single genomes (see Gadagkar et al. (2005)). Datasets corresponding to the three genomes (nuclear, mitochondrial, and plastid) were used individually and collectively to generate a molecular phylogeny of some apicomplexan parasites including the haemosporinids (Perkins et al. 2007), some coccidia in the family Sarcocystidae (Votýpka et al. 1998; Slapeta et al. 2003; Monteiro et al. 2007), and some Eimeria spp. (Zhao et al. 2001; Zhao and Duszynski 2001).

In the present study, the trees generated from 18S rDNA sequences, both as a single gene tree using all near-complete 18S rDNA sequences available for parasite in the Eimeriidae or as a subset of taxa represented across all three genomes, supported the monophyly of individual species (where more than one sequence was available for a single parasite) using taxa in the Sarcocystidae as the taxonomic outgroup. However, the paraphyly of the genus Eimeria (see Tenter et al. 2002; Morrison et al. 2004; Whipps et al. 2012) was clearly demonstrated in these phylogenies because clades of apparently closely related Eimeria spp. (often from closely related definitive hosts) were interspersed with parasites belonging to other genera such as Lankesterella, Isospora, Atoxoplasma, and Cyclospora. The occurrence of Atoxoplasma and Isospora spp. of avian origin in the same clade confirms previous studies (Carreno and Barta 1999; Barta et al. 2005; Schrenzel et al. 2005), suggesting that these coccidia of birds are closely related. The early branching of Eimeria arnyi and E. ranae, infecting colubrid snakes and frogs, respectively, is in agreement with previous analyses (e.g., Jirků et al. 2009; Whipps et al. 2012). The placement of Goussia spp. and Eimeria tropidura (found in amphibians and reptiles, respectively) supports the hypothesis that suture-bearing (rather than Stieda body-bearing) sporocysts are a symplesiomorphic character shared across many coccidial groups (see Jirků et al. 2002; Whipps et al. 2012) including the adeleorinid coccidia; in contrast, there is a single monophyletic clade of coccidia (most Eimeria spp., Cyclospora spp., Caryospora spp.) that all possess Stieda bodies in their sporocysts rather than sutures (Fig. 1). Our large nu 18S rDNA sequence-based analysis corroborates several studies suggesting that the Stieda body of coccidian sporocysts is a reliable, derived morphological trait that indicates membership in the eimeriid clade of parasites (Carreno and Barta 1999; Tenter et al. 2002; Jirků et al. 2009). Lankesterella spp. (and presumably closely related Schellackia spp. as well) are hypothesized to have secondarily lost this morphological feature during their evolution from monoxenous parasites of the digestive tract of poikilotherms to heteroxenous hemoparasites (albeit with only circulating sporozoites within the vertebrate blood and retention of vertebrates as their definitive hosts).

The alignments and resulting phylogenetic reconstructions generated from global and a subset of COI sequences had markedly smaller intraspecific genetic distances and variation compared with analyses based on nuclear 18S rDNA sequences. This appears to be a useful feature of mitochondrial genes for both species delimitation and phylogenetic resolution of the evolutionary history of apicomplexan parasites as suggested by Ogedengbe et al. (2011) and Hikosaka et al. (2010).

The phylogenetic trees generated from concatenated plastid gene sequences reflected, in large part, parasite/host coevolution. Eimeria trichosuri, isolated from the possum Trichosurus cunninghami (see Power et al. (2009)), branched early from Eimeria from birds and mammals.

A total evidence tree was generated from a concatenation of the nuclear 18S rDNA, mitochondrial cytochrome c oxidase I gene, and the concatenated plastid genes (rDNA and rpoB partial sequences). Eimeria from marsupials (E. trichosuri) and rodents (E. falciformis) formed a trichotomy with a monophyletic group that included all sampled Eimeria spp. from chickens (c.f. Power et al. 2009). Even though we did not have data from some of the Eimeria spp. found in rodents, the observed branching order was similar to those illustrated by Zhao and Duszynski (2001) who primarily used plastid ORF470 sequences and nuclear 18S rDNA sequences in separate analyses. Zhao et al. (2001) showed that a combined dataset including nuclear 18S rDNA and plastid 23S rDNA sequences could usefully delineate species that occur in rodents and bats.

From the foregoing, useful genes for molecular phylogenetic studies of Apicomplexa come from the nuclear and mitochondrial genomes. Additional information obtained from the plastid genome appears to be most useful as part of a concatenated dataset involving all three genomes and the five genes sequenced from them. The mitochondrial COI gene has the advantage of less intra-specific sequence variation within taxa, which has made this genetic locus a useful DNA barcoding target (Ogedengbe et al. 2011) that is free of the intraspecific variation observed for the 18S rRNA locus in coccidia (e.g., El-Sherry et al. 2013). Both mitochondrial and plastid genes appear to be maternally derived (Wilson and Williamson 1997; Ferguson et al. 2005) and are therefore less prone to genetic recombination resulting from sexual reproduction. The partial plastid genes used in the present study, while useful, are probably best utilized as concatenations (see Suchard et al. 2003; Gadagkar et al. 2005).

Nuclear rDNA sequences are widely used as genetic loci for apicomplexan parasites and many other eukaryotes. However, stage-specific rRNAs of nuclear origin that are differentially expressed in various life cycle stages have been demonstrated for some apicomplexan parasites (e.g., McCutchan et al. 1995; Li et al. 1997). This results in paralogous sequences within some apicomplexan taxa (through gene duplication and subsequent genetic changes) that can confound phylogenetic analyses based on rDNA sequences. Such paralogous sequences have been demonstrated in a number of Eimeria species as well (Vrba et al. 2011; El-Sherry et al. 2013), although unlike for Plasmodium spp. (see McCutchan et al. 1995; Li et al. 1997), the stage-specific expression of such rDNA paralogs has not been demonstrated in Eimeria species.

Studies using plastid or mitochondrial genes for phylogenetic analyses involving apicomplexan parasites are limited compared with those using nu 18S rDNA sequences. This study has made initial steps to address this information gap by generating novel sequences from the three genomes to compare their usefulness for molecular phylogenetic studies and evaluate their use in a multigenome/multigene analysis. Most phylogenetic studies of the family Eimeriidae, based mainly on nuclear genes (frequently nu 18S rDNA), show a pattern of parasite-definitive host coevolution, but these studies suffer from problems associated with using a single gene as noted above and as suggested by Martinsen et al. (2008). The observations in the present study indicate that a multigene approach could be useful in addressing both genomic and genetic evolution especially as it relates to parasite–host coevolution within and between members of the Apicomplexa that possess all three genomes, especially in addressing the polyphyly and paraphyly of economically important genera of parasites such as those found in the Eimeriidae (see Hikosaka et al. (2010)). Our suggestion would be that deeper evolutionary events may be best recovered using phylogenetic analyses based on nu 18S rDNA sequences and that mitochondrial COI sequences are highly informative at inferring more recent evolutionary events among closely related taxa. Nuclear 18S rDNA sequences combined with a rapidly growing collection of publicly available mitochondrial COI sequences together provide a compact and informative molecular dataset for inferring the evolutionary relationships among apicomplexan taxa.