Dear Hasan

For a combined analysis of 16S and COI markers, you should evaluate phylogenetic congruence among the two dataset partitions. It can be tested under the maximum parsimony (MP) criterion using the incongruent length difference (ILD) test implemented in the PAUP program as the partition homogeneity test.

If the topologies of the two data-sets were congruent (p >0.05), the sequences would combine.

I hope I answered your question

Best

You should either combine sequences in a single matrix (concatenation) or analyze them separately (species tree method) in *BEAST. I would recommend to do both. Geneious concatenates sequences pretty easily but it is not free. 

I agree with both answers above, but for small phylogenies a visual comparison might be good enough, comparing the phylogeny of both markers, and then a concatenated one.

A free alternative to concatenate the sequences is SequenceMatrix, if you have the collection number first it will recognize them by itself and concatenate them. It's also good for getting an overview if you have several markers. *BEAST might be a bit of an overkill if you're only making a phylogeny, MEGA has a way easier interface, and can do most analyses, and is free too(unlike geneious).

DAMBE is fine and free for concatenation and further analysis of mitochondrial genes. The software's link is http://dambe.bio.uottawa.ca/DAMBE/dambe.aspx

Hi!

You can concatenate the two matrix on Mesquite software (http://mesquiteproject.wikispaces.com/home). This software is a very easy platform available on Windows and Mac OS X. It is important to obtain a topology for COI and another one for 16S RNA, then to compare the topologies and to obtain a tree with both markers together.

These trees (for separate markers and both together) can be calculate by a distance method as NJ (on MEGA software) or by a phylogenetic method as Parsimony (TNT or PAUP software), Maximun Likelihood (on RaxmML and RaxMLGUI software) or Bayesian Inference (on Mr. Bayes software).

It is important to have a criterion for a substitution model in your sequences before to make your analysis. You can infer it on JModel test software.

In Youtube are a lot of tutorials for this kind of analysis. You can looking for this information for a exploratory aproximation.

*All software mentioned is free except PAUP.

Best regards from Mexico.

You  can concatenate  the COi and 16S  in to a single matrix in DAMBE, then you can use MEGA7 it has a easy interface. It is free, and easy for beginners and advanced analysis of  DNA and protein sequences.  In this software you can do a multi task such as  align the sequences with Clustal W or MUSCLE, find the nucleotide substitution model and  reconstruct the phylogeny based on distances like NJ , and others aproximations  like  ML and parsimony. One of the advantage in the distances methods such as NJ is that the result in a single tree.  MrBayes is another software that has a different approximation, it use Bayesian and MCMC that result in  a one consensus tree 

http://www.megasoftware.net/

http://mrbayes.sourceforge.net/