First of all, coalescent methods are useful if you have allelic variation in your data and manage to clone these alleles so they can be analysed phylogenetically. If you can't clone the alleles, for some reason, the aligment containing the "heterozygous" sequences *can not* be analysed phylogenetically but need to be displayed as, for instance, a splits network.

Also, you would be better off analysing a set of low-copy nuclear genes: both chloroplast markers and ITS are problematic to use for inferring a species phylogeny, at least at shallow taxonomic levels (5-10 Ma). ITS consists of thousands of near-identical copies in the genome, of which your PCR product will be a weighted average. In other words, if you detect "heterozygosity" for ITS, you cannot be sure whether it is because of paralogy or allelic variation. So even if you managed to clone your ITS copies, you would still not know what they represent. Chloroplast markers are usually uniparentally inherited and assumed not to recombine, which leads to different fitness among lineages and chloroplasts being more prone to introgression (chloroplast capture) than nuclear loci. Chloroplast markers are therefore expected to poorly reflect the species phylogeny.

Dear Renuka

Whenever performing phylogenetic analyses we mostly follow a representative approach. One of our assumptions then is, that our studied organisms are good representatives, usually of species (you may chose other taxonomic entities as operational taxonomic  Units, but let us take species as an example).

Each undetected intraspecific variation then represents a possible source of error (if it exceeds interspecific variation at least it does for sure, but generally it does, since model assumptions are usually violated).

Heterozygocity is just one special case of intraspecific variation. Heterozygocity often has the "Advantage" that it does not easily remain unnoticed (what you get is ambiguous results in PCR, or assembly if you perform NGS).

So yes it might be a Problem, but on the other hand, unlike other forms of intraspecific variability, it is often detected as a source of Problem.

Greetings

Nikola

There are several approaches to overcome the problem of heterozygous sites in sequences, none of them without methodological caveats

Or in order to deal with ambiguities, you can

1) calculate tree with double number of concatanated sequences, phased by random by DnaSP

2) if you have possible hybrids you can phase your sequences with potential parental species sequences and calculate tree

3) construct tree with RaxML and consider ambiguities as additional state or treat them as default - see RaxML manual

4) Phangorn package in R calculates genetic distance file with specific substitution/coding  scheme. (see ALASTAIR J. POTTS1,∗, TERRY A. HEDDERSON1, AND GUIDO W. GRIMM2 Constructing Phylogenies in the Presence Of Intra-Individual Site Polymorphisms (2ISPs) with a Focus on the Nuclear Ribosomal Cistron Syst. Biol. 63(1):1–16, 2014) Output file (Distnex.nex) could be processed in SplitsTree4 – Distance_network. Box-shape structures indicate on hybrid origin (Potts et al., 2014).

5) If you have many loci (tens or even thousands) coalescent methods could be complementary to traditional tree inference with ML, Bayesian or distance based methods

First of all, coalescent methods are useful if you have allelic variation in your data and manage to clone these alleles so they can be analysed phylogenetically. If you can't clone the alleles, for some reason, the aligment containing the "heterozygous" sequences *can not* be analysed phylogenetically but need to be displayed as, for instance, a splits network.

Also, you would be better off analysing a set of low-copy nuclear genes: both chloroplast markers and ITS are problematic to use for inferring a species phylogeny, at least at shallow taxonomic levels (5-10 Ma). ITS consists of thousands of near-identical copies in the genome, of which your PCR product will be a weighted average. In other words, if you detect "heterozygosity" for ITS, you cannot be sure whether it is because of paralogy or allelic variation. So even if you managed to clone your ITS copies, you would still not know what they represent. Chloroplast markers are usually uniparentally inherited and assumed not to recombine, which leads to different fitness among lineages and chloroplasts being more prone to introgression (chloroplast capture) than nuclear loci. Chloroplast markers are therefore expected to poorly reflect the species phylogeny.

Hi Renuka,

The points mentioned above are very much relevant when you are doing molecular phylogentic analysis and encounter 'parlog' type sequences. For ITS, there are a lot of published papers for its use in molecular phyogentic analysis as well as some papers on the problems associated with its use. In general the ITS region is part  for nuclear rRNA gene unit regions and like many other multi-copy sequences these undergo extensive  sequence homogenization (also as it has structural and functiona constraints). Generally, whether a plant is diploid (or higher ploidy) if the ITS sequence is homogenized, you may not find paralogs in the genome and the representative sequence is used for analysis.

However, people have used many other regions also for such analysis and generally it is advised to use multiple loci for inferring  phylogenetic relations ships. Data from one loci may not be perfect always.

One should always keep in mind that a dendrogarm derived from a sequence is anway a 'gene tree' and it may not always match well with the 'species tree'

best regards

Ajay Saini 

you guys are awesome. thank you all for your replies. I was out in the field for several days and could not be in tauch

Thanks a lot and these are very informative..

Hi Renuka,

Here are two papers will help you out in this respect.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2493047/

http://mbe.oxfordjournals.org/content/11/3/426.full.pdf

Good luck,

Dear All.. what a discussion group. I am thankful to all of you.. this is such a cool place where isolation by distance is broken dawn....:)

thank you very much all...

regards

renuka