In my experience, NJ trees are a good start point. It is very easy to construct a NJ tree with MEGA and the results are usually indicative of the phylogeny you are looking for. I suggest you to use the simple p-distance as model/method, it means that your will be clustered in function of their similarities. However, depending on what you want to do with your data, NJ is usually too much basic. You should also try to construct a ML tree and/or a Bayesian tree. These are accepted by the scientific community as robust methods, even if the open questions about the best strategy to use in phylogeny are still a lot. There are several programs to infer phylogeny using ML an Bayesian. For ML, try to use Garli (very good but not so much friendly) or Phyml. The last is also a command line program, but some services on the web allow to perform the analyses for you using external servers (e.g. http://www.atgc-montpellier.fr/phyml/) and are very easy to use, just paste your alignment (pay attention to the alignment, it is a fundamental step to obtain a reliable phylogeny). For Bayesian analyses, try MrBayes, it is very very good and, as already suggested you by Massimo Meregalli, it is very well supported by tutorial and manual. Both in ML and Bayesian, you need also to infer the best fitting substitution model to get the real (hopefully) phylogeny. MEGA has the function Models, you can try to start with this. But try also jmodeltest, it is not too much difficult to use and a little better. Good luck with your research.

model-- you can run jmodeltest or modeltest first to decide which model is more suitable for your data.

Bootstrap -- usually is 1000, sometime 500 is acceptable.

NJ -- NJ or ML, you need make decision based you data, your object and time consuming

You could also try a Bayesian inference. You can download the program MrBayes, and there is a very good tutorial and manual, with also examples.

You can try 1000 replicates for bootstrap analysis.

In my experience, NJ trees are a good start point. It is very easy to construct a NJ tree with MEGA and the results are usually indicative of the phylogeny you are looking for. I suggest you to use the simple p-distance as model/method, it means that your will be clustered in function of their similarities. However, depending on what you want to do with your data, NJ is usually too much basic. You should also try to construct a ML tree and/or a Bayesian tree. These are accepted by the scientific community as robust methods, even if the open questions about the best strategy to use in phylogeny are still a lot. There are several programs to infer phylogeny using ML an Bayesian. For ML, try to use Garli (very good but not so much friendly) or Phyml. The last is also a command line program, but some services on the web allow to perform the analyses for you using external servers (e.g. http://www.atgc-montpellier.fr/phyml/) and are very easy to use, just paste your alignment (pay attention to the alignment, it is a fundamental step to obtain a reliable phylogeny). For Bayesian analyses, try MrBayes, it is very very good and, as already suggested you by Massimo Meregalli, it is very well supported by tutorial and manual. Both in ML and Bayesian, you need also to infer the best fitting substitution model to get the real (hopefully) phylogeny. MEGA has the function Models, you can try to start with this. But try also jmodeltest, it is not too much difficult to use and a little better. Good luck with your research.

For ML, you can also download a quite friendly program, raxmlGUI, available from http://sourceforge.net/projects/raxmlgui/ - It usually works pretty well. As indicated by Giorgio, what's important is to infer the best substitution models. But if you run both Bayesian and ML, you can then check the resulting phylogenies. When the data are good, and the models are properly selected, you should obtain very similar phylogenies, with usually some differences in support of nodes (Bayesian often being higher), but with the same overall topology. Consider that the theoretical bases for Bayesian and ML are different, so if you obtain similar results, it's a good indication that your reconstruction has a high probability of being the best possible. You may need to read a bit about the theoretical background of ML and Bayesian, there are several good approaches on the web. Good luck!

You could, as well, try to use Maximum Parsimony. POY would be a good software to try it and it also allow you to use ML and BI.

My two cents: you may want to play around with more powerful tools than MEGA, at least if your dataset grows.

Here's a not-overly-long guide:

https://www.researchgate.net/publication/235601668_Incorporating_molecular_data_in_fungal_systematics_a_guide_for_aspiringresearchers?ev=prf_pub

Article Incorporating molecular data in fungal systematics: A guide ...

Hi!

Regarding the outgroup: The perfect outgroup to set the root to your tree is the closest relative to the stem group of the group under study. I also was told that the outgroup behaves more robustly, if two outgroup taxa, an inner one very close to the group under study, and an outer one a little further away are included. As I see that your samples all come from the Galloanserae, the pig seems a little far away, and I could imagine the analysis would benefit from inserting some outgroup taxa from Sauropsida. For example, ostrich and alligator would be my first choice. If you can not lay your hands on the actual animals, I wouldn't hesitate to search GenBank for suitable sequences (there might be some, as both taxa have economical value...)

regards

Paavo

MEGA 6 is a user-friendly program that allows us to perform sequence alignment, find the best model, and construct a phylogenetic tree (e.g. ML, NJ, etc.). The program can be downloaded from the following website:

http://www.megasoftware.net/