The Amblyomma maculatum Koch, 1844 (Acari: Ixodidae) group of ticks: phenotypic plasticity or incipient speciation?

Background The goal of this study was to reassess the taxonomic status of A. maculatum, A. triste and A. tigrinum by phylogenetic analysis of five molecular markers [four mitochondrial: 12S rDNA, 16S rDNA, the control region (DL) and cytochrome c oxidase 1 (cox1), and one nuclear: ribosomal intergenic transcribed spacer 2 (ITS2)]. In addition, the phenotypic diversity of adult ticks identified as A. maculatum and A. triste from geographically distinct populations was thoroughly re-examined. Results Microscopic examination identified four putative morphotypes distinguishable by disjunct geographical ranges, but very scant fixed characters. Analysis of the separated mitochondrial datasets mostly resulted in conflicting tree topologies. Nuclear gene sequences were almost identical throughout the geographical ranges of the two species, suggesting a very recent, almost explosive radiation of the terminal operational taxonomic units. Analysis of concatenated molecular datasets was more informative and indicated that, although genetically very close to the A. maculatum - A. triste lineage, A. tigrinum was a monophyletic separate entity. Within the A. maculatum - A. triste cluster, three main clades were supported. The two morphotypes, corresponding to the western North American and eastern North American populations, consistently grouped in a single monophyletic clade with many shared mitochondrial sequences among ticks of the two areas. Ticks from the two remaining morphotypes, south-eastern South America and Peruvian, corresponded to two distinct clades. Conclusions Given the paucity of morphological characters, the minimal genetic distance separating morphotypes, and more importantly the fact that two morphotypes are genetically indistinguishable, our data suggest that A. maculatum and A. triste should be synonymized and that morphological differences merely reflect very recent local adaptation to distinct environments in taxa that might be undergoing the first steps of speciation but have yet to complete lineage sorting. Nonetheless, future investigations using more sensitive nuclear markers and/or crossbreeding experiments might reveal the occurrence of very rapid speciation events in this group of taxa. Tentative node dating revealed that the A. tigrinum and A. maculatum - A. triste clades split about 2 Mya, while the A. maculatum - A.triste cluster radiated no earlier than 700,000 years ago. Electronic supplementary material The online version of this article (10.1186/s13071-018-3186-9) contains supplementary material, which is available to authorized users.


Background
The Amblyomma maculatum group includes, according to Camicas et al. [1], the following species: A. maculatum Koch, 1844; Amblyomma neumanni Ribaga, 1902; Amblyomma parvitarsum Neumann, 1901; Amblyomma tigrinum Koch, 1844 and Amblyomma triste Koch, 1844. Together with the Amblyomma ovale Koch, 1844 group which encompasses A. ovale and Amblyomma aureolatum (Pallas, 1772), they have been clustered by Camicas et al. [1] in a revised version of subgenus Anastosiella, originally erected by Santos Dias [2], who also included within this subgenus Amblyomma pecarium Dunn, 1933 and Amblyomma brasiliense Aragão, 1908. However, within the A. maculatum group, the adult and immature stages of A. neumanni and A. parvitarsum are morphologically distinguishable from the other species. Unlike the other three taxa, they are both characterized by incomplete marginal grooves in males, with A. parvitarsum having beady and orbited eyes [3]. In females, aside from A. neumanni, all species are glabrous dorsally. Amblyomma parvitarsum females also have beady and orbited eyes. Other diagnostic differences are listed in Estrada-Peña et al. [3], who suggested that A. neumanni and A. parvitarsum should be grouped with the A. ovale group in a yet to be determined subgenus, with A. maculatum, A. triste and A. tigrinum the only remaining members of the subgenus Anastosiella.
Members of the subgenus Anastosiella are morphologically very similar. This similarity is more apparent between A. maculatum and A. triste, and is examined in depth in this study. Koch [4] briefly described the three taxa based on males of A. maculatum and A. tigrinum, and a female of A. triste. He completed his description in 1850 [5] and essentially reported differences in punctation and ornamentation. Neumann [6] synonymyzed A. tigrinum and A. triste with A. maculatum after failing to observe differences in the shape of distal spines (modified setae) on tibiae II to IV (called tarsi by [6], protarsi by [7] and metatarsi by [8]). Kohls [8] reestablished A. tigrinum and A. triste as valid species and completely redescribed the three taxa. In that work, the characters differentiating the three species were the presence/absence of tubercles on the festoons, and the presence of one or two spurs on "metatarsi" of legs II-IV. Nevertheless, the identification chiefly of A. maculatum and A. triste has consistently been challenging [3,[9][10][11] and marred by frequent misidentifications [12][13][14]. Taxonomic issues are not limited to adult stages: immatures, for which taxonomic keys are available, are even more difficult to differentiate [3,11,[15][16][17] with, again, the exception of A. parvitarsum and A. neumanni which are easily separated from the other three species.
The distribution of A. maculatum is presumably confined to the southern USA, Central America and some areas of Colombia, Venezuela, Peru and Ecuador, whereas A. tigrinum is reported to occur only in South American countries [8,[18][19][20][21]. Amblyomma triste was regarded as being exclusively South American in its distribution until recently, when it was reported from Mexico and the southwestern USA [11,22], thus increasing the number of tick species with both, a Neotropical and Nearctic distribution [9].
Molecular techniques used to infer phylogenetic relationships and evaluate the taxonomic status of the different species of the A. maculatum group have yet to be applied in a comprehensive manner. Preliminary reports based on the analysis of 16S rDNA sequences confirmed that A. maculatum, A. triste and A. tigrinum are closely related to each other, whereas A. neumanni and A. parvitarsum are distinct not only from each other and from the rest of the A. maculatum group of taxa, but do not cluster with the A. ovale group of species [3].
The aim of this study was to reassess the taxonomic status of A. maculatum, A. triste and A. tigrinum by analyzing their phylogenetic relationships determined by comparisons of one nuclear and four mitochondrial gene sequences. In addition, a comprehensive morphological analysis of the adult stage of A. maculatum and A. triste is presented. Our working hypothesis is that A. maculatum, A. triste and A. tigrinum comprise three separate species.

Results
Morphological reassessment of the A. maculatum -A. triste specimens Microscopic examination identified four morphological groups that, for the sake of simplicity, will be called here Morphotypes I, II, III and IV. The four morphotypes have disjunct geographical distributions ( Fig. 1; morphotype distribution for the samples used in the molecular analyses) and feature many common and a few distinctive character combinations ( Table 1). The shape of modified setae on the ventral distal end of the tibiae of legs II-IV (tibial armature) have been commonly used to differentiate ticks of the A. maculatum group. They have been called spines, setae or spurs in earlier literature. In the following descriptions we will use the term "seta" if the structure is fine and weakly sclerotized, "spine" if the structure is thick and heavily sclerotized. For measurements, refer to scales on the images. Body outline elongate-oval, narrower anterior to eyes; scapulae rounded; eyes flat. Scutum ornate, with reddish brown spots outlined by pale yellowish enameled stripes; postero-median spot extending anteriorly to level of spiracular plates; postero-accessory spots parallel to postero-median spot; three lateral spots large and sometimes fused; cervical spots narrow, first diverging, then slightly converging posteriorly; central area long, extending to mid-length of scutum. Chitinous scutes sometimes reduced to a tubercle, but always present on ventral surface of festoons. Basis capituli dorsally sub-rectangular, cornua long. Hypostome spatulate, dental formula 3:3. Legs: trochanters without spurs, coxa I with two distinct spurs, external spur long and sharp, internal spur as a small tubercle; coxae II-III with triangular, short, blunt spur each; coxa IV with narrow long sharp spur.
Female. Brown central spot extending to the posterior margin of the scutum present (Fig. 2d). One spine and one seta on tibiae of legs II-IV present as in male. Spiracular plates usually oval, with short, wide dorsal projection longer than in males, dorsal projection as wide as adjoining festoon (Fig. 2g).  Male. Chitinous scutes present on ventral surface of festoons, often less salient than in Morphotype I, sometimes reduced to a posteromedian tubercle (Fig. 3c-e). Two robust spines almost equal in size present on tibiae of legs II-IV (Fig. 3f). Spiracular plates comma-shaped, with tip of dorsal projection equal in width or slightly narrower than half of adjacent festoon; dorsal projection perpendicular to longitudinal axis of spiracular plate (Fig. 3e). Spurs on coxa IV not reaching anus (Fig. 3c); postero-median spot wider than enameled stripe between postero-median and postero-lateral spots (Fig. 3a, b).
Female. Two spines almost equal in size present on tibiae of legs II-IV (Fig. 3g). Spiracular plates comma-shaped, with dorsal projection perpendicular to axis of spiracular plate, as wide or slightly wider than half-festoon width (Fig. 4h). Scutal brown central area long and narrow, not reaching posterior margin of scutum in most specimens, a few specimens with central area reaching posterior margin of the scutum; enameled surface of scutum variable in shape and proportions (Fig. 4a-f).  Description (Fig. 5a-h) Male. Chitinous scutes present on ventral surface of festoons (Fig. 5g), visible dorsally (Fig. 5a-c). Postero-median spot wider than enameled stripe between postero-median and postero-lateral spots in most specimens (Fig. 5c), a few specimens with postero-median spot almost equal in width to stripe between postero-median and postero-lateral spots. One spine and one seta present on tibiae of legs II-IV (Fig. 5e). Spiracular plates comma-shaped, with dorsal projection approximately half as wide as adjoining festoon; dorsal projection as an extension of spiracular plate, not perpendicular to its axis (Fig. 5b). Spurs on coxa IV not reaching anus (Fig. 5g).
Female. Brown scutal central area long and narrow, always reaching posterior margin of scutum (Fig. 5d). One spine and one seta present on tibiae of legs II-IV as in male. Spiracular plates with dorsal projection perpendicular to plate axis, as wide or narrower than a third of adjoining festoon (Fig. 5a).  Male. Chitinous scutes present on ventral surface of festoons (Fig. 6e), visible dorsally (Fig. 6b). Two spines slightly different in size present on tibiae of legs II-IV but originating at a different level along tarsi and therefore appearing to be of different lengths (Fig. 6d). Spiracular plates comma-shaped, with dorsal projection about half as wide as adjoining festoon extending from plate at an obtuse angle (Fig.  6i). Postero-median spot wider than enameled stripe between postero-median and postero-lateral spots (Fig. 6b).
Female. Central scutal area long, narrow (Fig. 6a), reaching posterior margin of scutum in most specimens. Spines on tibiae of legs II-IV as in males (Fig. 6g). Spiracular plates comma-shaped, with dorsal projection perpendicular to plates and approximately half as wide as adjoining festoon.

PCR amplification and sequence alignment
Although we did not obtain all gene sequences for each sample, we obtained sequences representative for all species and morphotypes (   Figure 1 illustrates the sites where the ticks used for phylogenetic analyses were collected.

Sequences and haplotype diversity
If we exclude the outgroup sequences, the 87 12S rDNA gene sequences were represented by 17 unique haplotypes (Additional file 1:

Phylogenetic analyses Separate datasets
The separate gene sequence analyses (phylogenies not shown) were characterized by overall limited resolution. The ingroup was consistently strongly supported. Amblyomma tigrinum was always monophyletic, mostly as the sister lineage to the A. maculatum -A. triste cluster (16S rDNA, DL and ITS2), but also as the sister group of Peruvian Morphotype IV (cox1) or embedded within Morphotype I (12S rDNA). The overall structure of the five phylogenies was different, with Morphotypes II and III never clustering in two separate lineages, while the southern South American and the Peruvian sequences were sometimes identifiable as distinct clades (cox1). In addition, the analysis of ITS2 sequences only identified a supported polytomic ingroup. A closer examination of the ITS2 sequences revealed that when the matrix included outgroups, informative characters were 108 (11%); however, when outgroup sequences were excluded, informative characters only represented 0.9%. In the ITS2 ingroup matrix substitutions were mostly randomly scattered singleton mutations and indels with no phylogenetic information.
Mitochondrial concatenated datasets (12S rDNA + 16S rDNA + cox1 + DL) The MP (Maximum Parsimony) analysis of the mtDNA matrix found 6586 equally parsimonious trees (length = 506). The resolution of the MP strict consensus tree was better than that recovered for the separate datasets. The A. tigrinum and the A. maculatum -A. triste clusters were sister clades and both were well supported. Within the A. maculatum -A. triste clade, the first split occurred between Morphotype IV (Peru) and the remaining lineages. Morphotypes II and III were intermixed within a single clade, except for one sequence (identified as CO-US/GA in the tree) which corresponded to a Colombian sequence. Morphotype I was sister to Morphotypes II+III and included all southern South American sequences. Bayesian analysis recovered an almost identical topology (Fig. 7). Within these supported clades, resolution was very limited and when monophyletic, clades did not correspond to any geographical or ecologically meaningful subset.
Mitochondrial-nuclear concatenated dataset (12S rDNA + 16S rDNA + cox1 + DL + ITS2) The concatenated dataset represented a matrix of 2556 bp (31 sequences, corresponding to 27 unique haplotypes and one outgroup). The MP analysis identified a total of 56 equally parsimonious trees (length = 770). The MP and Bayesian inference BI reconstructions were almost completely congruent: the ingroup was monophyletic and A. tigrinum was consistently identified as the supported sister clade to the monophyletic A. maculatum -A. triste clade (Fig. 7b). However, by MP, the A. maculatum -A. triste clade was polytomic leading to monophyletic Morphotypes I, II+III, and IV. While Morphotype II was found to be monophyletic, Morphotype III did not cluster in a single supported clade. By BI, Morphotype IV was found to be the sister clade to the remaining morphotypes which clustered in a well-supported lineage (Fig. 7b). Morphotype I was monophyletic and, again, the structure within Morphotype I did not appear to correspond to geographical subdivisions with sequences from different regions represented in each clade. The last monophyletic clade included a monophyletic Morphotype II and a paraphyletic Morphotype III (Fig. 7b). Divergence values in the mt+nDNA dataset, calculated after the best mutation model was evaluated by ModelTest, showed that the distance between outgroup and ingroup varied from 18.60 to 19.01%, between A. tigrinum and A. maculatum -A. triste by 3.91-4.31%, between Morphotype IV and the remaining clades by 1.22-1.70%, between Morphotype I and II+III by 0.91-1.22% and between Morphotype II and III by 0.32-0.71%. Within Morphotype II, distances varied between 0 and 0.28%, within Morphotype III between 0 and 0.51% and within Morphotype I between 1.22 and 1.70%.

Node dating
DAMBE analyses revealed no significant mutation rate differences between the main ingroup lineages (P = 0.053-0.9) including A. tigrinum. Both the least-square method and the likelihood ratio test did not reject the molecular clock hypothesis, further confirming that the lineages under consideration did not evolve at significantly different rates. Although this would justify the use Fig. 7 Phylogenetic reconstruction of the mitochondrial concatenated dataset (a), and the total evidence (mitochondrial + nuclear) concatenated dataset (b). Trees were inferred by Bayesian analysis. Numbers on branches represent maximum parsimony bootstrap support and Bayesian posterior probability support (in bold) of a strict molecular clock in BEAST, we selected the relaxed molecular clock option which allows for variable mutation rates because mutation rates differed significantly between ingroup and outgroup (P ≤ 0.01). The tree generated by BEAST supported the basal subdivision between A. tigrinum and A. maculatum -A. triste, but did not resolve relationships within the A. maculatum -A. triste polytomic clade (Fig. 8). By using a single calibration point, we first confirmed that the nodes within the A. cajennense group were consistent with prior findings [36] associating nodes with known biogeographical vicariant events. The diversification of the A. maculatum group of taxa appeared to be more recent with the split between A. tigrinum and the A. maculatum -A. triste clades dating back to approximately 2.1 Mya (95% confidence interval: 0.9-3.3 Mya) and the A. maculatum -A. triste radiation beginning no sooner than 720,000 years ago (95% confidence interval: 0.3-1.2 Mya) in the mid-Pleistocene (Fig. 8). The split between the outgroup and the ingroup dated was estimated at 39 Mya (95% confidence interval: 14-70 Mya).

Discussion
Morphological reassessment of the A. maculatum -A. triste specimens revealed four morphotypes distinguishable by four distinct sets of characters. Morphotype I includes specimens belonging to populations from Argentina, Paraguay, Uruguay and southern Brazil.
These ticks were assigned to the taxon A. triste (s.s.) because their morphology is compatible with the morphology of the type-specimens of A. triste (ZMB 1046) from Montevideo, Uruguay (see [4]). Morphotype II includes specimens from eastern USA, Colombia, Guatemala, Honduras, Mexico and Venezuela, with a morphology matching the type-specimens of A. maculatum (s.s.) (ZMB 1044) also described by Koch [4] from a type-locality reported as "Carolina", USA (possibly in one of the Carolina States which are included in the present distribution of A. maculatum). Morphotypes I and II have so far been differentiated by examining the tibial armature of legs II-IV, the shape of the male spiracular plates and, in some cases, scutal ornamentation in females. The remaining two morphotypes present a combination of characters that precludes a strict assignation to either A. triste or A. maculatum, which explain why Peruvian and Chilean samples have alternatively been considered to be A. maculatum or A. triste [10,37,38], and Arizona samples have been assigned to A. triste [11], based on tibial armature morphology. Morphotype III includes ticks from the southwestern USA (Texas and Arizona) and northern Mexico. It could also correspond to what Guzman-Cornejo et al. [22] described as A. triste from Mexico, although the specimens should be reexamined. They have one spine and one seta on tibiae of legs II-IV, comma-shaped spiracular plates (elongated in males), and the scutum of females with a central spot reaching the posterior margin of the scutum. Morphotypes III and IV are similar, but Morphotype IV is characterized by the presence of two spines which appear to be of different length on tibiae of legs II-IV. However, when these phenotypic differences are closely scrutinized, with the exceptions of the overall shape of the spiracular plates and differences in thickness of the tibial setae on legs II-IV, none of the other phenotypic features appear to be fixed when sufficient numbers of ticks from each area are observed. All other differentiating features are shared by at least two of the four morphotypes (Table 2). This situation raises the question of whether we are dealing with four different species, less than four, or a single species.
Analyses of the individual gene sequences, commonly used to reassess taxonomic issues within Amblyomma taxa [36,[39][40][41], were unable to resolve relationships within the group. Only analyses of the concatenated datasets could subdivide the A. maculatum -A. triste group into three supported clades characterized by distinct geographical ranges, but not always by distinct morphologies since Morphotype III is not monophyletic but is instead a basal polytomy within the Morphotype II+III cluster. The genetic similarity, often identity, particularly between mitochondrial sequences from eastern USA and Arizona-Mexico (Additional files 1, 2, 3, 4: Tables S1-S4) indicate that the spurs and the features of the spiracular plate commonly used to distinguish A. maculatum from A. triste are not species-specific characters. Molecular divergence values between morphotypes II and III are equivalent to diversity within Morphotype I, further indicating that the two morphologically distinct but genetically uniform North American populations, can be considered to belong to the same species, A. maculatum, which extends into parts of northern South America. Although we only obtained sequences from a single Colombian sample of A. maculatum, it consistently clustered with a sample from Georgia, USA (21D) as a slightly separated lineage within the Morphotype II+III clade, confirming the occurrence of the species in Colombia, as recently shown in a study on the diversity of Colombian ticks [21].
If the morphology of the tibial armature and of the spiracular plates cannot consistently be associated with speciation, the next question is whether we should consider Morphotype I, i.e. A. triste (s.s.), and Morphotype IV to be different species from A. maculatum. Genetic distance between populations of ticks that are geographically remote or even disjunct in their distribution is to be expected. Nevertheless, even the largest distances within the A.maculatum -A. triste group are comparable to intraspecific distances in other Amblyomma [36,39,40] and Ixodes [42][43][44] species when corresponding gene distances are compared separately. Nevertheless, mutation rates can vary between more or less related lineages and comparisons of simple distance values between distant taxa can be misleading. In this case, however, we have verified that within the A. maculatum group mutation rates are not significantly different between the sister lineages. Because distance values within Morphotype I (0.91-1.22%) are equal to or higher than distances between the same Morphotypes I and II+III (0.32-0.71%), we deduce that A. triste should also be returned to a junior synonym of A. maculatum (priority by page number in [4]). Although distances between Morphotype IV and the other morphotypes are slightly higher (1.22-1.70%), they are significantly lower than those between A. tigrinum and all morphotypes (3.91-4.31%).
In this study we are facing a situation in which morphological and molecular data are not offering a congruent solution, in a taxon (the ingroup) that radiated very recently. Our data do not support consistently the subdivision of this group into four species, although we agree that some of the presented evidence would provide justification. For now, we can only base our conclusion on our results, but there is no doubt that additional studies at a finer taxonomic level based either on nuclear codominant very variable markers (microsatellites or SNPs) and cross-breeding experiments will help in reaching more satisfying conclusions.
In the case of A. tigrinum, this species differs from A. triste and A. maculatum by host preferences [17,[45][46][47], ecological preferences [17] and morphology of both adults and immature stages [17]. Molecularly, A. tigrinum sequences always clustered in a well-defined monophyletic lineage. Although divergence values between A. tigrinum and A. maculatum -A. triste are moderately higher than intraspecific values, they remain much lower than the interspecific distances recorded between outgroup species, and between outgroup and ingroup taxa. More importantly, the variable nuclear gene used in this study (ITS2) which has successfully been used for taxonomic reassessments among South American Amblyomma species [36,41], like some of the mitochondrial genes, included A. tigrinum in a polytomic ingroup and did not support a clear split between the taxa. For the time being, we do not consider A. tigrinum to be a synonym of A. maculatum, as proposed by Neumann [6] but to be a separate taxon of very recent radiation from A. maculatum -A. triste. After all, it has been shown that in some cases differences in host association can trigger rapid genetic divergence in tick species [48]. Only cross-breeding experiments are likely to determine whether the time elapsed since the divergence of the two lineages (2.1 Mya) has been sufficient for them to become different species.
In terms of node dating, the A. maculatum -A. triste radiation initiated no earlier than 700,000 years ago, a time frame not too different from that evaluated for the intraspecific diversification of A. variegatum [39] and of Ixodes scapularis Say [43,44]. The branch lengths within the A. maculatum -A. triste clade are very short compared to those between the ingroup and outgroup and between different outgroup species (Fig. 8). This indicates that the radiation from the earliest common ancestor of the A. maculatum group was very rapid, almost explosive. If we compare our tree topology to that of other Amblyomma groups (the A. cajennense and A. parvum groups) with similar geographical distributions [36,40], it appears that the A. maculatum group colonized its present range much later and very rapidly. Alternatively, we might consider the possibility that earlier widespread populations went almost extinct and that the present lineages arose through a major bottleneck. It could be argued that Morphotypes II-III-IV have evolved along the same biogeographical patterns described for A. mixtum [36] a species which exploited the closing of the Panama Isthmus (3 Mya) for dispersal through a wide area encompassing Colombia, the coast of Ecuador, Central America and southern North America. This could speculatively have been followed by allopatric incipient diversification occurring along the western coast of northern South America (Peru), in the Madrean Archipelago (Sky Islands) north of Mexico and in southern Arizona [49], while the most common morphotype (II) survived as such in Colombia and southeastern North America. Vicariant events cannot, however, explain the split between Morphotype II and I.
Additional studies with different molecular markers and many more samples, could help to better reconstruct the demographic history of the group.

Conclusions
In summary, the data and evidence presented are inconsistent with the hypothesis that A. maculatum, A. triste and A. tigrinum represent three separate species. The evidence presented in this study supports the conspecificity of A. maculatum and A. triste. It is possible that the minor observed morphological differences are the result of a very rapid adaptation to slightly different environments not yet associated with sufficient genetic differentiation to support speciation. Further studies, especially cross-breeding experiments, should follow, as they may add valuable information and further support (or reject) the hypothesis of conspecificity of A. maculatum and A. triste raised in this study.

Morphological analysis
Morphological analyses were performed through examination of adult ticks deposited in the following tick collections: ( [17].

Sampling for molecular analyses
Our sample included a total of 109 adult specimens initially identified as A. maculatum, A. triste or A. tigrinum, and two specimens identified as A. parvitarsum and A. neumanni. Amblyomma maculatum ticks were from the USA (Georgia and Florida), Peru and Colombia; A. triste were from Argentina, Brazil, Uruguay, Peru, Mexico and Arizona (USA); A. tigrinum were from Argentina and Brazil; and A. parvitarsum and A. neumanni were from Argentina. When available, specimens from several localities were included to consider variation between and within different regions ( Table 1). The ticks were subsequently reclassified by using a newly outlined phenotypic subdivision (see Results section) of A. maculatum -A. triste into four morphotypes (Table 1). Ticks were obtained from 28 localities and 7 countries (Fig. 1 (Table 1).

DNA extraction, PCR and sequencing
Tick DNA was extracted and, when possible, the exoskeletons were preserved for further morphological analysis following previously published protocols [39,50]. A small portion of the postero-lateral idiosoma of each tick was removed by using a sterile disposable scalpel and the tick was incubated overnight in 180 μl Qiagen ATL lysis buffer (Qiagen, Valencia, CA, USA) and 40 μl of a 14.3 mg/ml solution of proteinase K (Roche Applied Sciences, Indianapolis, IN, USA). After repeated vortexing and ascertaining that the lysis was complete, each exoskeleton was stored in 70% ethanol and kept as a voucher specimen. The lysed tissues were further processed as previously described [36,39,50]. Five mitochondrial gene sequences, 12S rDNA (small subunit ribosomal RNA), 16S rDNA (small subunit ribosomal RNA), cox1 (cythochrome c oxidase subunit 1), the control region or d-loop (DL) were amplified with previously reported sets of primers [36,39,[50][51][52]. In addition, a portion of the nuclear ribosomal internal transcribed spacer 2 (ITS2) was amplified by slightly modifying a previously published protocol to include 35 instead of 27 annealing cycles [39,53]. PCRs were performed using a MasterTaq kit (5-Prime, Gaithersburg, MD, USA). Each reaction contained 2.5 μl of tick DNA, 2.5 μl of 10× Taq buffer, 5 μl of 5× TaqMaster PCR Enhancer, 1.5 μl of MgAc (25 mM), 0.5 μl of dNTP mix (10 mM each), 0.1 μl of Taq polymerase (5U/μl), 1.25 μl of each primer from a 10 pmol/μl stock solution (Invitrogen, Life Technologies Corporation, Grand Island, NY, USA) and 14.6 μl of molecular biology grade water. The two DNA strands of each amplicon were purified and sequenced at the High-Throughput Genomics Unit (HTGU, University of Washington, Seattle, WA, USA or at Eurofins, Louisville, KY, USA) and were assembled with Sequencer 4.5 (Gene Codes Corporation, Ann Arbor, MI, USA).

Phylogenetic analyses
Sequences were manually aligned with McClade 4.07 OSX (Sinauer Associates, Sunderland, MA, USA) [54]. Secondary structure was considered in aligning ribosomal genes [50] and DL [55]. Codon organization was taken into account when aligning the cox1 data set. Each data set was analyzed by maximum parsimony (MP) with PAUP [56] and by Bayesian inference analysis (BI) using MrBayes 3.2.4 [57,58]. Branch support was assessed by bootstrap analysis (1000 replicates) with PAUP for MP, and by posterior probability with MrBayes for BI. MP heuristic searches were performed by branch-swapping using the tree bisection-reconnection (TBR) algorithm. Maximum likelihood distances were calculated after the nucleotide substitution model best fitting the data was selected by JModeltest v.2.1.7 [59,60]. Maximum likelihood pairwise distances were calculated based in the selected model by using PAUP. Two runs with four chains each were run simultaneously for BI analyses (1,000,000 generations). Trees were sampled every 100 iterations. Trees saved before the average standard deviation of split fragments converged to a value < 0.01 were discarded from the final sample. When necessary, the number of generations was increased so that the number of discarded samples would not exceed 25% of the total sampled trees. The 50% majority-rule consensus tree of the remaining trees was inferred, and posterior probabilities recorded for each branch. Two concatenated datasets, one (mtDNA) including 4 mitochondrial genes (12S rDNA, 16S rDNA, DL and cox1) and one including the same 4 gene sequences and ITS2 sequences (mt+nDNA), were analyzed following the same procedure outlined for the separate analyses. A. parvitarsum and A. neumanni were used as outgroups in our analyses because they are recognized as being close relatives of our ingroup taxa [3]. Additional species were considered as possible outgroups and preliminary analyses were performed with the following taxa: A. aureolatum; Amblyomma coelebs Neumann, 1899; Amblyomma dubitatum Neumann, 1899; Amblyomma oblongoguttatum Koch, 1844; and A. ovale. These alternative outgroups were discarded because they were too distantly related to the ingroup to provide resolution within ingroup taxa.