What Do Pneumocystis Organisms Tell Us about the Phylogeography of Their Hosts? The Case of the Woodmouse Apodemus sylvaticus in Continental Europe and Western Mediterranean Islands

Pneumocystis fungi represent a highly diversified biological group with numerous species, which display a strong host-specificity suggesting a long co-speciation process. In the present study, the presence and genetic diversity of Pneumocystis organisms was investigated in 203 lung samples from woodmice (Apodemus sylvaticus) collected on western continental Europe and Mediterranean islands. The presence of Pneumocystis DNA was assessed by nested PCR at both large and small mitochondrial subunit (mtLSU and mtSSU) rRNA loci. Direct sequencing of nested PCR products demonstrated a very high variability among woodmouse-derived Pneumocystis organisms with a total number of 30 distinct combined mtLSU and mtSSU sequence types. However, the genetic divergence among these sequence types was very low (up to 3.87%) and the presence of several Pneumocystis species within Apodemus sylvaticus was considered unlikely. The analysis of the genetic structure of woodmouse-derived Pneumocystis revealed two distinct groups. The first one comprised Pneumocystis from woodmice collected in continental Spain, France and Balearic islands. The second one included Pneumocystis from woodmice collected in continental Italy, Corsica and Sicily. These two genetic groups were in accordance with the two lineages currently described within the host species Apodemus sylvaticus. Pneumocystis organisms are emerging as powerful tools for phylogeographic studies in mammals.


Introduction
Phylogeography is a field of research, which studies the principles and processes determining the geographical distribution of genetic lineages. Quite frequently, parasites provide an additional source of information and can better reconstruct the common history of hosts and parasites [1]. For this reason, some parasites have been recently used as phylogeographic markers [1][2][3][4][5][6][7]. As an example, host associations of Spinturnicid mites likely result from a complex interaction between the phylogenetic history of the bat hosts and the behaviour and the ecology of both parasites and hosts [6]. Moreover, Spinturnicid mites seem to be a biological tag to reveal the former presence of a nowadays locally extinct bat host species [7]. Similarly, studies on the co-divergence between A. sylvaticus and the parasitic nematode Heligmosomoides polygyrus [2,8] demonstrated a general association between the parasite genetic structure and the host evolutionary history.
The woodmouse (Apodemus sylvaticus, Rodentia, Muridae) is a common forest dweller, and has been present throughout Europe for at least 3 million years [23]. Apodemus sylvaticus is one of the most intensively studied species in the genus. The phylogeographic history of this species in Europe and in the Middle East is well known, proposing an original pattern of postglacial colonization where the Iberian and southern France refuge populations colonized almost all European regions [23][24].
In the present study, we detected and characterised Pneumocystis organisms in woodmice from western continental Europe and Mediterranean islands. Questions regarding host and parasite co-divergence as well as comparative phylogeography were specifically addressed. Our objective was to assess whether Pneumocystis organisms can be used as phylogeographic markers for rodents, in addition to previous studies on parasites of Apodemus and in order to demonstrate that Pneumocystis is a promising biological tag for mammalian hosts.

Samples
Woodmice were trapped on islands and mainland areas in western Europe (Fig. 1

DNA extraction
DNA extraction from lung tissue samples was performed using DNeasy tissue kit (Qiagen, Courtaboeuf, France) following the manufacturer procedure with some modifications. Lysis of lung tissue was done overnight in an incubator at 56°C with permanent rotation. In order to concentrate Pneumocystis DNA, the column was re-hydrated with 100 μl of elution AE buffer instead of 200 μl. Extracted DNA was stored at -20°C prior to amplification. A negative control was systematically included in each series of DNA extraction.

Detection of Pneumocystis organisms by nested PCR
Pneumocystis DNA was detected in pulmonary samples by nested PCR at the mitochondrial large subunit (mtLSU) rRNA and the mitochondrial small subunit (mtSSU) rRNA genes. At the mtLSU rRNA locus, the first round of nested PCR consisted of 30 cycles and was performed using the external primer set PH/PE [25]. Each cycle consisted of 3 steps: denaturation for 30s at 94°C, annealing for 1 min at 50°C, and extension for 1 min at 72°C. Five percent (vol/vol) of the amplified product from the first PCR round was used as starting matrix in the second round of PCR. The second PCR round consisted of 30 cycles and was performed using the internal primer set PX/PZ [20]. Each cycle consisted of 3 steps: denaturation for 30s at 94°C, annealing for 1 min at 55°C, and extension for 2 min at 72°C.
At the mtSSU rRNA locus, the first round of nested PCR consisted of 40 cycles and was performed using the external primer set pAZ112-10F-R1 / pAZ112-10R-R1 [26]. Each cycle consisted of 3 steps: denaturation for 30 s at 94°C, annealing for 1 min at 55°C, and extension for 1 min at 65°C. Five percent (vol/vol) of the amplified product from the first round of PCR was used as starting matrix in the second round. The second round of nested PCR consisted of 40 cycles and was performed using the internal primer set pAZ112-13-R1 / pAZ112-14-R1 [26]. Each cycle consisted of 3 steps: denaturation for 30 s at 94°C, annealing for 1 min at 52°C, and extension for 1 min at 65°C for 10 cycles; and denaturation for 30 s at 94°C, annealing for 1 min at 63°C, and extension for 1 min at 65°C for 30 cycles.
Negative controls were included in each experiment, in both DNA extraction and PCR amplification. When non-specific bands were detected, amplification products of expected size (about 250-300 pb) were extracted from a 2% agarose gel (run in Tris-borate-EDTA buffer) using a PCR purification kit (QIAEX II Gel Extraction Kit, Qiagen, Courtaboeuf, France). Purified products were sent for sequencing to GenoScreen (Pasteur campus, Genopole of Lille, France). When a unique band of expected size was present, amplified products were directly sent for purification and sequencing to GenoScreen. Sequencing from both ends using sets of internal primers was performed on an automated DNA sequencer (3730XL DNA Analyser, Applied Biosystems). Not all the positive samples were sequenced; representative positive samples were randomly chosen to be sequenced.

Nucleotide sequence accession numbers
Pneumocystis mtLSU rDNA or mtSSU rDNA sequences obtained in this study were deposited in GenBank under accession numbers KF411466, KF384955 to KF385004, and KF384913 to KF384954.

Statistical analysis
Statistical analyses were performed using SAS (SAS v9, SAS Institute). Variables are described as counts (proportions). A Mc Nemar test was performed to assess the agreement between detection of Pneumocystis DNA using mtLSU and mtSSU rRNA. Pneumocystis infection was investigated using a two-way analysis of variance with country and insularity as crossed fixedeffect factors, analyzed with a generalized linear mixed model approach (proc GLIMMIX). Models were built with a binomial error distribution and the logit link function. Two-sided p-values<0.05 were deemed to be statistically significant.

Sequence analysis
The mtLSU and mtSSU rRNA gene sequences described in this study were concatenated and initially aligned with homologous sequences of Pneumocystis from the mouse Mus musculus (P. murina GenBank accession numbers AF257179 and KF384954, respectively) and from the squirrel Sciureus aestuans (GenBank accession numbers KF411466 and KF384953, respectively) using CLUSTAL X v1.63b [27]. The sequence alignment was then refined by visual optimization using Se-Al v2.0a11 [28] respecting the four criteria defined by Barriel [29]: (1) minimize the number of mutations, which means substitution and insertion-deletion events, and prefer substitution to insertion-deletion; (2) favour transitions to transversions because they are more frequent; (3) minimize the number of sites affected by mutation; (4) minimize the number of phylogenetic implications. All analyses were performed using sequences from P. murina and Pneumocystis of the squirrel S. aestuans as outgroups. Once the alignment was finalised, the insertion-deletion events were coded as "id" to exploit their phylogenetic information, independently of their size, following the procedure of Barriel [29]. This procedure is particularly useful for complex zones with complex insertions/deletions with substitutions. This strategy requires the introduction of three new character states in the matrix: "i" (insertion), "d" (deletion) and question marks "?" as missing data.
Maximum parsimony (MP) analysis was conducted using Phylogenetic Analysis Using Parcimony Program (PAUP) v4.0b9 [30]. A MP tree was inferred using a heuristic search of 1,000 random additions of sequences with TBR (tree bisection reconnection) branch-swapping. All sites were equally weighed, gaps were scored as missing (the ID code was included in the matrix, so gaps were taken into account).
In a second step, the positions corresponding to missing data in some sequences were removed from the original alignment. Maximum likelihood (ML) analysis of sequences was carried out with PHYML v3.0 [31] using the GTR (general time reversible) + Γ (gamma distribution of rates with four rate categories) + I (proportion of invariant sites) model. The appropriate model of sequence evolution was selected using the program jModelTest2 [32]. Bootstrap values (BVs) were obtained from 1,000 pseudo-replicates. Bayesian analysis was also performed with the program MrBAYES v3.1 [33] on the same data set using the same model and Bayesian Posterior Probabilities (BPPs) were calculated using a Markov Chain Monte Carlo (MCMC) sampling approach [34]. The starting tree was random, and four simultaneous Markov chains were run in duplicates for 2 million generations with burn-in values set at 10,000 generations (based on empirical values of stabilizing likelihoods). Trees were sampled every 100 generations (the other settings being set to default). Distance matrices and bootstrap sampling (1,000 pseudo-replicates) were subsequently generated by using Seqboot and Dnadist combination (Phylip v.3.69) with the Jukes Cantor model. Matrices were analysed using BioNJ [35] and consensus tree computed with Consense (Phylip v.3.69). Alignments of mtLSU and mtSSU rRNA gene sequences utilised in the various analyses are available upon request to the corresponding author.

Phylogeographic analyses
A minimum spanning network was constructed using the MINSPNET algorithm available in the ARLEQUIN 2.0 program [36]. The genetic divergences (GD) between sample groups were estimated using a distance analysis (K 2 P, mega program). The nucleotide (π) and haplotype (h) diversities were estimated using the DNA sequence polymorphism (DnaSP) version 5 program [37].

Detection of Pneumocystis DNA in woodmice lungs
A sample was considered to be positive for the presence of Pneumocystis DNA when a specific band was amplified by PCR at the expected size either at the mtLSU or mtSSU rRNA locus or at both loci (Table 1). Pneumocystis DNA was detected in 189 out of 203 woodmice lung samples examined (93.1%). Amplifications at both mtSSU and mtLSU rDNA were positive for the same individual in 38 animals (Table 1).

Potential influence of country and insularity versus mainland in Pneumocystis carriage
No significant statistical association could be established between the presence of Pneumocystis DNA and the country of origin of animals (p = 0.79) ( Table 2). No interaction was observed between country and insularity (p = 0.85). However, the frequency of Pneumocystis DNA detection was significantly higher in woodmice from continent than in insular woodmice (p = 0.02). The probability of picking a Pneumocystis-infected woodmouse was 6-fold higher on the continent than on islands (Table 2).

Sequence analysis
Direct sequencing of nested PCR products demonstrated a very high variability among woodmice-derived Pneumocystis sequences. A total number of 32 distinct mtLSU sequence types and 13 distinct mtSSU sequence types were detected. Sequences at the mtLSU rRNA locus seemed to be more variable than those at the mtSSU rRNA locus (Table 1). After combining mtLSU and mtSSU sequences, 30 distinct sequence types were generated (Tables 1 and 3). Phylogenetic analyses were then carried out using the concatenated sequence dataset. A total number of 555 bp (360 bp and 195 bp from mtSSUrRNA and mtLSUrRNA loci, respectively) was selected in the subsequent phylogenetic analyses. The initial alignment included 32 taxa (2 outgroups and 30 taxa ingroups), 555 characters and 513 common sites. After coding the insertion-deletion events [29], the final alignment with the ID code included 32 taxa and 584 characters. Genetic diversity Most of the combined sequences corresponded to specific geographical areas (Tables 1; Fig. 2). Three combined Pneumocystis sequence types were detected in the woodmice from the Pyrenees (France). Similarly, unique Pneumocystis sequence types were characterized in the  Table 1) was common with 2 other ones isolated in woodmice from Montseny (Spain, Spa19) and Porquerolles (France, Por15). Among the 3 specific Pneumocystis sequence-types found in the woodmice from Calabria (Italy), one Pneumocystis sequence was common with that found in a woodmouse from Corsica (France). In addition, 2 unique Pneumocystis sequence-types were detected in the corsican woodmice ( Table 1). The total genetic divergence among the studied localities (Table 3) delineated 2 groups of combined sequence types. The first one, named the Western European clade, comprised Pneumocystis from woodmice collected in continental Spain (Montseny), France and Balearic islands (range of genetic divergence: 0% to 0.2%). The second one, named the Italo-Corsican clade, included Pneumocystis from woodmice collected in continental Italy (Calabria), Corsica and Sicily (range of genetic divergence: 0% to 0.3%). The genetic divergence existing between the Western-European and Italo-Corsican clades ranged from 0.8% to 1.4% and is higher than the genetic divergence existing within each clade (Table 3).
When comparing the genetic diversity values of Pneumocystis isolated from woodmice captured on islands versus those from host populations originating from mainland, a trend toward the loss of Pneumocystis genetic diversity was noticed. Indeed, the nucleotide diversity tends to decrease (π) in the Pneumocystis sequences isolated from insular host populations except for Sicily (8 individuals and 8 sequences types) and Minorca (2 individuals and 2 sequences types) ( Table 4). With regard to both Pneumocystis groups ( Fig. 2 and 3), the sole insular Pneumocystis population that seemed to display levels of genetic divergence (GD) and mean nucleotide diversity (π) that were comparable to those assessed in population from continental Italy (Calabria) was isolated from Sicilian rodents. All other Pneumocystis populations from Mediterranean islands were characterised by lower values of GD and π than mainland source populations except for Minorca ( Table 4).
The Pneumocystis continental populations from Calabria, Montpellier and the Pyrenees displayed the highest levels of π except for Montseny (Table 4).    Geographic origins (see Table 1 and Fig. 1)  The analysis of the genetic structure of woodmice-derived Pneumocystis revealed two distinct groups (Fig. 2). The Pneumocystis clade from Italy and Corsica was well individualized in the phylogenetic analyses (NJ, MP, ML, MB) and it was supported by bootstrap values higher than 80%. A small subgroup corresponding to most of the Pneumocystis sequences isolated from woodmice trapped in Sicily (Sc3, Sc8, Sc9, Sc13 and Sc16), was weakly or well supported by bootstrap values depending on the various analyses applied (Fig. 2).
Pneumocystis sequence types from Western European woodmice (Porquerolles and mainland France, Balearic Islands and mainland Spain) always met outside of the Italo-Corsican clade and consequently constituted a second group of Pneumocystis that was not well supported by bootstrap values and was not monophyletic (Fig. 2). The relationships existing inside this group remain unclear with the exception of 2 Pneumocystis sequence types isolated from 2 woodmice trapped in Montpellier (Mon23 and Mon4) (Fig. 2).

Phylogeographic analyses
The minimum spanning network displayed 24 haplotypes of Pneumocystis clearly distributed in 2 major clades defined above and separated from each other by a genetic distance of 5 mutational steps (Fig. 3). However, two other clades seemed to appear, each constituted by only one Pneumocystis haplotype isolated from 2 woodmice trapped in Montpellier (Mon23 and Mon4). These Pneumocystis types were clearly divergent from the others. For the 2 major clades, the first one comprised Pneumocystis haplotypes resulting from woodmice from Sicily, Calabria and Corsica (clade 1). The second group included Pneumocystis haplotypes from woodmice from Spain, Montpelier, Mallorca and Minorca as well as the Pyrenees (Western European clade 2). Within clade 2, almost all haplotypes were connected together in a star-like topology. Pneumocystis isolated from woodmice captured in Spain, Montpellier, Porquerolles and the Pyrenees (Spa19, Mon5, Por15, Spa4 and Pyr14) were grouped into the same haplotype. Another Pneumocystis sequence-type isolated from woodmice sampled in Spain fell into another haplotype shared with a sample from Minorca (Spa3 and Mn33).
The topology of the Italo-Corsican clade (clade 1) is more heterogeneous than in the Western European clade. We note the proximity between the Pneumocystis from the woodmice from Corsica, Calabria and Sicily, but also 2 haplotypes shared by Pneumocystis from Calabrian and Corsican woodmice (Fig. 3). On the other hand, Sicilian woodmice possessed their own Pneumocystis haplotypes and did not share them with the Corsican or Calabrian woodmice (Fig. 3).

Prevalence of Pneumocystis in Apodemus sylvaticus
Previous studies using nested PCR reported important differences in Pneumocystis prevalence between mammal species of the same zoological group such as primates or rodents. Prevalence of Pneumocystis reached 26.5% in captive primates [38] and 33.6% in healthy macaques maintained in partial release [39]. In micromammals collected in France, Pneumocystis prevalence reached 68.0% when molecular techniques were applied [40]. In Thailand, 57.7% of wild rats harboured Pneumocystis [41]. Pneumocystis was found in rodents of the genus Apodemus in several European and Japanese regions, with rates of prevalence ranging from 0 to 43% [42].
The prevalence reported in the present study (93.1%) was much higher than previously cited values. This result may be explained both by the efficiency of molecular amplification used for Pneumocystis detection in our study, and by the gregarious behaviour of A. sylvaticus [43] as crowding favours the transmission of Pneumocystis by aerial route [19,39,44]. The Pneumocystis transmission is influenced by the interactions between mammals and the structuring of mammalian society [19,39]. Consequently, Pneumocystis can be considered as a marker of social structuring of mammals.

Potential influence of insularity versus continent
The prevalence of Pneumocystis colonization was significantly higher in woodmice from continent than in the ones from islands. This finding is not in accordance with previous study [45]. Indeed, the prevalence and intensity of parasitism are often higher in insular than in mainland populations for parasites that are directly transmitted from host-to-host. This high insularity prevalence is in relation with the low parasite species richness in insular host populations. A study found that there was a significant influence of insularity on the decrease in parasite species richness of A. sylvaticus [46], which can be explained by the decrease of mammal diversity [47].
Another study on H. polygyrus and its host A. sylvaticus highlighted the "island syndrome" in this parasitic nematode species with a loss of genetic diversity following territory colonization [48]. Our results tend to support this observation (Table 4). In the first clade (Western European Pneumocystis woodmice), the Majorcan Pneumocystis population displayed a significant loss of genetic diversity (GD = 0.1%; π = 0.0011). Even if nucleotidic diversity of the Minorcan Pneumocystis population appeared to be higher than the diversity of the Spanish population (mainland origin), both were yet lower than those of Montpellier and the Pyrenees. Moreover, Nieberding et al. [48] observed in the case of Minorcan H. polygyrus that the GD and π values were intermediate when compared to those from Spain and Majorcan populations. The lower values of GD and π of Pneumocystis Spanish populations seem to reflect past genetic bottlenecks or recent population expansion from a small number of founder individuals.
In the case of the second clade (Calabria, Sicily and Corsica), we observed the same tendency of decrease in Pneumocystis genetic and nucleotidic diversity in islands, except for Sicily, thus confirming the results obtained previously for A. sylvaticus and its nematode parasite H. polygyrus [23,48].
The results of the present work support the hypothesis that the Pneumocystis Balearic populations originated from South-Western Europe and the Italian region was the continental source for Pneumocystis Sicilian and Corsican populations as well as for its host A. sylvaticus and other parasites like H. polygyrus [48].

Genetic diversity of Pneumocystis from Apodemus sylvaticus
Several studies reported Pneumocystis sequence types from various host species. The genetic divergence ranged from 7.5% to 28.2% according to the used locus, i.e. mtLSU rRNA or mtSSU rRNA, and the host species examined [10,13,19]. The genetic divergence among Pneumocystis organisms from A. sylvaticus was very low (lower than 3.87% of divergence) and strongly suggested that a single Pneumocystis species is present in the woodmouse. The phylogenetic analysis of Pneumocystis sequences was in accordance with the host specificity concept of Pneumocystis [14][15]17]. A first investigation about primate-derived Pneumocystis organisms demonstrated that Pneumocystis phylogeny mirrors its host phylogeny, suggesting long-range functional and genetic adaptation process leading to cospeciation [21]. Pneumocystis organisms may have evolved together with their hosts. The low genetic divergence in woodmice-derived Pneumocystis organisms is in connection with the geographical place where the rodents were captured.
A phylogenetic tree for A. sylvaticus was first provided by Michaux et al. [23,[49][50]. According to this analysis, two groups of populations can be distinguished: a first group comprises woodmice from France, Spain as well as Balearic Islands (Western European group); the second group includes woodmice from the Balkans, Italy and Sicily. The phylogenetic study inferred from Pneumocystis sequences detected in woodmice showed the same tendency of grouping (Fig. 2). Particularly, Pneumocystis from woodmice in Italy, Corsica and Sicily were clustered in the same monophyletic group. In contrast, the other Western European populations of Pneumocystis appeared less structured. Concerning the Mediterranean islands, our results pointed out a close relationship of the Porquerolles and Mallorca islands with the continental French and Spanish Pneumocystis populations. In contrast, Corsican Pneumocystis appeared strongly associated to the Italian and Sicilian Pneumocystis populations ( Fig. 2 and 3). Michaux et al. [50] demonstrated that A. sylvaticus underwent a significant bottleneck during Quaternary glaciations, finding refugia in the Iberian Peninsula and the Italo-Balkan areas. While the Iberian populations later spread throughout the rest of Europe, the Italo-Balkan populations remained trapped by surrounding mountains and did not contribute to the genetic constitution of central and north European populations. The results of the study suggest co-divergence: the population structure of woodmice-related Pneumocystis is in agreement with the evolutionary history of A. sylvaticus. Pneumocystis organisms were likely carried by their specific host during their migration and evolution.
In a recent study, using a nematode parasite as phylogeographic marker, Callejon et al. [51] underlined two clear-cut geographical and genetic lineages of the nematode Trichuris muris isolated from different Muridae hosts, including A. sylvaticus, in Europe: one lineage was widespread from Northern Spain (Catalonia) to Denmark (Western European region), while the second one colonized the Eastern European region (Croatia, Rumania, and Turkey) [51]. The geographic repartition of Pneumocystis sequences is thus in accordance with the historical processes of Apodemus speciation described and emphasized by the parasite population evolution.
A focus on the last clade, comprising woodmice from the Balkans, Italy and Sicily, pointed out a division into two subclades showing a large and robust differentiation [23]. Animals from Sicily displayed a higher level of nucleotide diversity than animals from Balkans and continental Italy [23]. Nieberding et al. [8] established the phylogeographical pattern of the European nematode H. polygyrus, and, like its woodmouse host, the Italo-Balkan lineage comprised a Sicilian subclade. Moreover, Nieberding et al. [48] reported that the southern Italian region is the continental source for Sicilian populations of H. polygyrus as well as for its host A. sylvaticus. In the present study, the phylogenetic tree and the minimum spanning network showed a comparable trend to the grouping of woodmice-related Pneumocystis in this Sicilian clade, even if some relations remain unclear. Sicilian Pneumocystis population displayed levels of GD and π that were comparable to Calabrian source populations. Moreover, many more haplotypes were found in Sicilian Pneumocystis than fungi from other regions as depicted by the network analysis. This suggests a more heterogeneous distribution of Sicilian Pneumocystis and thus, a more stable population less subjected to a recent expansion. These results tend to confirm the hypothesis in which Sicily would be a 'hotspot' of genetic diversity for woodmice and its associated parasites [48]. Some uncertainties remain in the Pneumocystis life cycle. Particularly, a question is unanswered: are Pneumocystis organisms able to survive or multiply in the environment? A recent study reports that the social or behavioural factors may influence transmission of Pneumocystis within the colonies of bat species, while the environmental factors do not seem to have an impact on the carriage rate of Pneumocystis [11]. The strong host-species specificity and co-divergence do not support the hypothesis of an environmental reservoir [2,6,[10][11]13]. This rather suggests that Pneumocystis may only survive very ephemerally in the environment, and that the infection travels from host-to-host via the airborne route, while the crowding of mammals favours fungal transmission. Pneumocystis could be used as a tool to study the social structuring and phylogeny of their hosts.
This double feature of Pneumocystis organisms, host and site specificities, constitutes a promising tool to unveil the phylogeny or the phylogeography for many mammalian hosts.

Conclusion
Pneumocystis species seem to be good candidates for phylogenetic and phylogeographic studies in Apodemus sylvaticus. Indeed, the prevalence of Pneumocystis carriage in A. sylvaticus is higher (93.1%) than that reported for other parasites like the nematode Heligmosomoides polygyrus also studied in A. sylvaticus. Thus, Pneumocystis organisms constitute an easily accessible marker. Furthermore, we determined host and parasite co-divergence. In woodmice-derived Pneumocystis, specific DNA sequence divergence among Pneumocystis genotypes was correlated with the geographic location of their specific hosts. The analysis of Pneumocystis genetic diversity confirmed the evolutionary history of A. sylvaticus in Europe. Furthermore, Sicily was confirmed to be a hotspot of intraspecific biodiversity not only at the level of the hosts but also of their parasites. Pneumocystis species and genotypes are emerging as powerful tools to conduct phylogeographic studies in mammals.
Supporting Information S1 Table. Divergence matrix of Pneumocystis combined mtLSU rDNA and mtSSU rDNA sequences amplified from 30 woodmice, two rodents (P. murina from mouse Mus musculus and Pneumocystis from squirrel Sciureus aestuans).