Results of this study revealed a high level of similarity among the mt genomes of F. circinatum, F. verticillioides and F. fujikuroi. Many of the shared characters observed in these genomes are also common among the mt genomes of other Sordariomycetes [18–22, 41]. This is evident from the overall synteny of the genomes, particularly with respect to gene order conservation of the nad2-nad3, the nad4L-nad5, the cob-cox1, the nad1-nad4, the atp8-atp6 genes, rns-trn-cox3-trn and nad6-trn genes.The mt tRNA genes of all Sordariomycetes examined so far are clustered and we also observed the characteristic conservation of gene order. Additional properties typical of the mt genomes of Sordariomycetes include transcription from the same strand, nad4L and nad5 genes overlapping by one nucleotide and biases towards TAA and ATG in stop and start codons, respectively. Although the mt genomes of Sordariomycetes are generally diverse with regards to intron density and type, these genomes are all characterized by the presence of a group I intron after the central loop of domain V in the rnl gene, which harbours an ORF encoding rps3. In agreement to what is typically found in other Sordariomycetes, the intergenic regions of Fusarium mt genomes were also highly diverse with much of the inter-specific sequence dissimilarity attributable to these regions.
The major tRNA gene cluster identified in the F. circinatum, F. verticillioides and F. fujikuroi mt genomes are interrupted by a unique ORF. The presence of this ORF could be a unique characteristic of all Fusarium species as it was recently also found in F. solani and F. graminearum. This suggests that the ORF could have been acquired by the most recent common ancestor of these species, possibly via a horizontal gene transfer event [49, 50]. The current analysis of codon usage also supported the notion that this ORF does not share the same ancestry as the rest of the genes on the mt genomes of these fungi, primarily because its codon usage differed from those of the known protein coding genes . Although this unusual codon bias could indicate that the ORF represents a pseudogene , the results of Al-Reedy et. al. confirming that this ORF is transcribed and encodes a membrane associated protein, suggest otherwise.
The sizes of the mt genomes sequenced for isolates of F. circinatum, F. verticillioides and F. fujikuroi differed considerably. The mt genome sequence of F. circinatum is approximately 13 323 and 19 350 bases larger than those of F. verticillioides and F. fujikuroi, respectively. These size differences were primarily due to the presence of group I self-splicing introns. This is because the exclusion of intron sequences resulted in mt genomes sizes that did not differ by more than 4100 bp (some of which reflect the presence of unique intergenic repeats). This variation in intron number and size was also observed in the F. oxysporum, F. graminearum and F. solani mt genomes. However, PCR analysis of the introns located in the cox1 gene of a small set of isolates of F. circinatum and F. verticillioides indicated that some introns were absent from certain individuals, while other individuals harboured additional introns not present in the sequenced isolates, which suggests that the mt genome sizes reported here are not fixed for each species. Although the involvement of potential PCR artifacts cannot be excluded, our results is consistent with what was shown previously for species within the genus Leptographuim.
The results of analyses of group I introns suggest ancestry involving both horizontal and vertical acquisition for these elements in the GFC [53–57]. The large diversity of introns in the mt genomes of F. circinatum, F. verticillioides and F. fujikuroi and the other Fusarium species outside of this complex  indicates that the acquisition of these elements occurred multiple times and independently [53, 55, 56]. In contrast, the fact that introns at comparable positions in genes appear to be ‘orthologous’, or at least encoding sequences sharing a high degree of similarity, suggests vertical transmission or inheritance from a common ancestor with subsequent loss in specific species and/or lineages [54–56]. The apparent conservation of certain intron positions in mt genes could be that these positions are more favorable for intron insertion (i.e., linked to the internal guide sequence) . Intron insertion position conservation within fungal mt genomes  and even across kingdoms [57, 58] have also been observed in recent studies.
Phylogenetic comparison of the LAGLIDADG and GIY-YIG domains identified in the HEGs of F. circinatum, F. verticillioides and F. fujikuroi and the other Fusarium mt genomes  reflected the known diversity of the HEGs encoding them . These genes are believed to have invaded group I introns with little association between intron class and HEG type and/or family [46, 60]. The LAGLIDADG and GIY-YIG phylogeny not only reflected the diversity of these HEGs in terms of multiple family members, but potentially also reflects the evolution of these elements as “selfish genes” that have acquired nonsense or frameshift mutations once they became fixed [61, 62]. Diversity within the HEGs identified from this study could therefore reflect non-functional HEGs, since the cycle of intron-without-HEG to intron-with–functional-HEG to intron-with-nonfunctional-HEG and ultimately back to intron-without-HEG is ongoing [61, 62]. However, to determine whether the phylogenetic diversity observed for the LAGLIDADG and GIY-YIG sequences examined here are due to decay through the acquisition of mutations, future experimental analyses would need to consider the functionality (i.e., mobility) of the HEGs within these mt genomes.
The second major objective of this study was to determine whether mt genes represent suitable DNA markers for tracing the evolutionary history of Fusarium species in the GFC. To accomplish this, we firstly compared the concatenated phylogeny inferred from the 14 mt protein coding genes with phylogenies inferred from nuclear genes described previously [25–30]. In the concatenated gene tree, F. fujikuroi represented the sister group of a F. circinatum + F. verticillioides clade, which suggests that the “Asian” Clade is ancestral to the “American” + “African” Clades. Although this grouping is in agreement with some reports [e.g. [28, 29], it contradicts previous notions that the “African” Clade is the ancestral group of the GFC. The “African” origins hypothesis for the GFC is based on the fact that the “African” Clade is most species rich and diverse and it also includes most mycotoxin producers [23, 25, 26]. More importantly, this Clade of the GFC is also the only one that includes chlamydospore-forming species, which is a characteristic thought to be common to Fusarium species outside of this complex [23, 25, 26]. Resolution of the evolutionary history of the GFC thus requires more work involving phylogenetic analyses of multiple additional gene regions and/or phylogenomic approaches using the information from whole nuclear genomes.
Next we compared individual mitochondrial gene tree topologies to the topology of the mitochondrial concatenated dataset. The working hypothesis was that the entire mt genome will reflect or support the same genealogical history , because it is thought to represent a single replicative unit that is inherited in a maternal fashion . However, at least six distinct phylogenetic trees were recovered from the various single-gene datasets. Furthermore, three of the single-gene did not allow the recovery of a monophyletic GFC (i.e., F. circinatum, F. verticillioides and F. fujikuroi were not grouped together into a single clade). These conflicts thus suggest that caution should be used when individual mt gene sequences are employed for phylogenetic analyses.
The phylogenetic incongruence among the different mt datasets that was observed in this study could be linked to the life history of the GFC. Previous authors have suggested that the origins of this complex are associated with a hybridization event, which gave rise to multiple non-orthologous copies of one of the internal transcribed spacer regions of the ribosomal RNA genes . If this were true, one would expect the different genes in the mt genomes of the species to be incongruent because of recombination that would have taken place among the genomes [31–33]. In fact, recombination among mt genomes is common when individuals are heteroplasmic due to bi-parental inheritance or parental leakage [43, 44], and parental leakage is typical during species hybridization . Although the observed incongruence among the mt gene trees in F. circinatum, F. verticillioides and F. fujikuroi support the notion of an ancient hybridization event, further research is needed to unambiguously resolve the early history of this important complex of fungal species.