This study focused on genetic diversity of G. duodenalis at the gdh gene using isolates collected from three different regions of Thailand. Cloning and sequencing approaches were used to elucidate heterologous alleles existed within the samples. Many studies have often detected overlapping nucleotide peaks which represented as mixed template at several genetic markers from different geographical locations . The result of mixed templates gives rise to a question whether this phenomenon is actually the result of mixed infection or the occurrence of ASH. Until now, there is still no direct evidence to prove which one plays a major role in the occurrence of ambiguous nucleotides. Thus, to provide conclusive evidence, further studies are required to explain the existence of ASH using cloned isolates of G. duodenalis which has never been shown by any studies. Although our study used the isolates from the patients without being cloned, to support the existence of ASH, indirect evidence of genetic exchange by recombination was obtained using bioinformatics studies.
The results obtained from the present study revealed that G. duodenalis isolates containing multiple alleles naturally presented in every area surveyed in Thailand, as shown by sequencing results of the subclones from isolates having overlapping chromatogram signals. These heterogenous sequencing results were observed only within assemblage B and throughout subtypes BIII and BIV whereas all assemblage A was homogeneous. The co-amplification of the cross-contaminated isolate was unlikely to occur because the isolates from each region were collected and processed at different times. Additionally, every isolate that revealed mixed templates was repeatedly tested under independent PCR and sequencing reactions. However, this finding seems to be common, as the occurrence of heterogeneous positions in the sequences of the gdh gene of assemblage A is markedly low . The presence of heterogenous nucleotides obtained from direct sequencing is usually considered to be the results of simultaneous infection with more than one Giardia assemblage. However, using the subcloning technique, the abundance of nine different gdh alleles observed in some isolates, lead us to presume that it could not be only the outcome of mixed infection. Hence, the existence of the ASH in these isolates should also be taken into consideration.
Alignment analysis of the polymorphic sites within assemblage B revealed that almost all nucleotide substitutions observed were synonymous changes, except for four positions. The Tajima's D test on the gdh gene showed contrasting results to those obtained with the β-giardin gene of other studies. The β-giardin gene was likely to be under the effects of ongoing purifying selection  while the gdh gene was under neutral selection. This suggested that molecular adaptation of these two genes might be influenced by different pressures. Furthermore, the computational prediction estimated that these changes did not influence the protein function. It indicated that variations appeared in the amino acid level are neutral or advantageous but not deleterious. The prediction is based on the non-structure method that considers the information from the amino acid sequence of interest, such as the position and type of amino acid changes, and compares their properties with the homolog protein family in the database . This method seems to be the most reliable option to predict the effect of the nonsynonymous substitution in this gene since most of the gdh gene studies are based on partial sequences. This may be due to the limitation of primer design to amplify the whole gene as this gene contains a number of variations and high percentage of GC content .
The estimation of the fixation index between three different sampling areas in Thailand did not support geographical sub-structuring within the G. duodenalis isolates. At present, the variations found in this study could not explain the geographical distribution of infected individuals. The only observation about the geographical aspect shown in this study is that the G. duodenalis populations were widely distributes throughout all three regions. The lack of geographical sub-structuring shown in this study was not unexpected since small fragments of only one gene were used to analyze the geographical distribution of this protozoan. Nevertheless, to the best of our knowledge, there is still no genotyping system that can efficiently indicate geographical sub-structuring of this organism, even using multilocus genes as genotypic markers . Whilst, the application of the high-resolution genotyping system is still necessary to address this question since it will be useful to distinguish different transmission routes and sources of infection.
Since the first finding of the genes known to function during meiosis and later confirmed by cloning and sequencing of PCR products [19, 38], the question about the potential capability of sexual reproduction in Giardia has been proposed. Subsequently, a number of studies have been conducted to provide evidence in support of genetic exchange among G. duodenalis isolates [18, 19, 39]. The present research attempted to extend the study of this issue to the next step by testing the potential of recombination events with the genetic data obtained from field isolates. In this study, we used the recombination analysis to show that the ASH could be a consequence of genetic exchange.
When the reticulate events, such as hybridization, gene transfer, and genetic recombination, are suspected to be involved, the phylogenetic network is one of the method that play a role in the accommodation of the non-treelike evolution. By using an agglomerative process implemented in the algorithm of Neighbor-Net, it can represent the conflicting signal or alternative phylogenetic histories, which are not adequately modeled by the bifurcating phylogenetic tree, in the format of a split graph. The presentation of the reticulations in the network indicated the possibility of interaction between two hypothetical ancestors to become extant taxa. In this study, the network tree clearly showed that the recombination might not be a phenomenon limited to laboratory strains and the interactions between taxa separately occurred within their own lineages of assemblages BIII and BIV.
Besides the evidence from the phylogenetic network tree, more intensive analyses were applied to further investigate the possibility of recombination from the dataset of this study. Two tests were selected based on their different assumptions for detecting the recombination to validate the evidence obtained from network tree. Four-gamete test is different from other general recombination testing methods that it is the population-specific method, generating to detect recombination between closely related genotypes. However, not all recombination events are revealed by this test due to its limitation that not support the occurrence of the recurrent or convergent mutations.
To confirm the results from the four-gamete test, a robust statistical test for recombination, Φ test, was applied. This recently developed approach is designed to operates under more relax model and has been proved through empirical data analysis that it can effectively discriminate between the presence and absence of recombination in both closely and distantly related samples .
The positivity of the four-gamete test and the statistical significance obtained from the Φ test strongly indicated the existence of the recombination in both subassemblages BIII and BIV. However, the recombination events were not significant when analyzing only sequence data of subassemblage BIV. This might be due to a small number of sequence data used for analysis (only 5 sequences tested). Low levels of variation among sequences limited the detection of recombination using this test . Generally, there are four major goals in the study of recombination that are i) detecting evidence of recombination in a dataset, ii) identifying the mosaic sequences, iii) delineating their breakpoints, and iv) quantifying recombination . Clearly, the majority of the Giardia studies, including this study, are in the early stage for recombination analysis that all evidences are indirectly detected from the mathematical and statistical models. Usually, if significant evidence for recombination can be detected, the localization of the recombination breakpoint is the next goal for the analysis. If the mosaic pattern of the sequence can be demonstrated, this will support the existence of genetic recombination in this organism.