If you want to use phylogeny to help resolving your species limits questions you will need one, or more outgroups to root the tree. You may alternatively not use phylogeny, and use other species limits tests, depending on how many genes you are using (GYMC, PTP, multiloci tests); you may consider looking at sequence divergence, describing discrete and testing quantitative morphology, etc. Of course you can also code morphological characters and add to your phylogeny - but again, if you use a phylogenetic approach you will need an outgroup for that matter.

It depends on your aims. If you have only 1 sequence for each species, outgroup is requred to esimate separatness. If you have number of sequences and not interested in phylogeny, outgroup is not required. But in case you want to see how close these taxa to each other, you must include in your sampling more than two species to see whole picture.

In the case you decide to use an outgroup. Try with several different, and see how it affect to your results.

Dear Andrii, thanks for the answer. Here I am having only one sequence (28S r DNA) for each species, So, according to you I have to add out group to compare the character states in the group of interest (the in-group) with those in a group that is closely related.

The outgroup will give you more insight to improve your tree results.

You should also try to get more sequence data from other species (GenBank e.g.) because using only two species to confirm these species seems not enough

The outgroup and from other sequence data from GeneBank will make your comparison and separation become very distinct.

If you want to test if these two species are distinct species, you will always have to add multiple specimens of both species to be able to make such a distinction. Even within a species you can get different sequences, so a distance between two samples is not enough. Adding outgroup taxa (best to add several, and at least the closest species if available) would also help to clarify this picture.

You should always have an outgroup sequence, and, as previously said, the outgroup can affect your results. You want a species close to the two that you are comparing to get the best results. Many time these sequences can be found in Genbank, or you'll have to generate your own. Good luck!

You only need an outgroup if you are interested in the order of divergences from a common ancestor: where is the first split and the second and so on or if you are interested in dating the tree. If you are simply asking how closely one taxa is to another or whether a group of organisms are monophyletic with respect to another group, then an unrooted tree is perfectly fine and has exactly the same information about topology and branch length and branch confidence.

In reply to Patrice: you are right that an unrooted tree will supply you with most information. But if you want to know if both entities (species) are monophyletic, you will need a rooted tree. The same unrooted tree (network) will result in two monophyletic entities when using one branch as the root, while it might result in one mono- and one paraphyletic entity when using another branch. Of course it also possible that the unrooted tree already clearly shows the whole thing is one big polyphyletic mess, and that it no longer matters where you put the root.

In reply to Jan J. Weiringa: I agree, but (if I understood correctly) Lalan Kumar Arya has only one sequence for each species. With that, I do not see much use on adding an outgroup. Obviously, adding more sequences for species and some outgroups will allow to address more questions.

Increased taxon sampling can only increase the confidence you will haven your results. Outgroup is a relative term. In this case I think you will want o sample as many closely related species within your 'Family' as possible.  Cheers, Adam 

Jan, You are right about needing at least one other clade other than those that are being checked for monophyly. But it need not be rooted. An unrooted tree is not a network. A network can have reticulations but a phylogenetic tree cannot. Note also that every tree search in every method, for computational reasons, infers an unrooted tree. We can and do draw it as if it is rooted and we certainly can add outgroup species to the search. But the resulting tree is always unrooted. We use outside evidence to place the root. Note that the likelihood score for the best tree(s) do not change with rooting.

The attached trees are both unrooted and both have the same monophyletic clades (a few are colored):

begin trees;

tree tree_1 = [&U] ((Aardvark:0.13766,(((Hemiechinus:0.07602,Hedgehog:0.09317):0.07166,Echinops:0.16712):0.0148,Elephant:0.1621):0.00315):0.00962,(((Rhinopholus:0.12215,Pteropus:0.12358):9.0E-4,((Pipistrellus:0.10839,Chalinolobus:0.09885):0.04547,FruitBat:0.12119):0.00749):0.00652,(((((Hippo:0.10932,(Sperm_Whale:0.09813,(FinWhale:0.04607,BlueWhale:0.03995):0.03922):0.02829):0.00905,(Cow:0.07949,Sheep:0.07113):0.02953):0.00262,Alpaca:0.13401):0.00254,Pig:0.11403):0.00447,(((Donkey:0.04264,Horse:0.04025):0.05039,(WhiteRhino:0.06368,IndiaRhino:0.0623):0.02573):0.01922,(Dog:0.12307,(Cat:0.10741,(GraySeal:0.01657,'Harb.Seal':0.01765):0.09089):0.00604):0.01523):9.0E-4):0.00675):0.00827,(Tupaia:0.15006,((Nycticebus:0.16311,(Cebus:0.14628,((Macaca:0.08603,Baboon:0.09037):0.04166,((Gibbon:0.0963,Orangutan:0.09331):0.00168,(Gorilla:0.06449,(Human:0.05198,Chimp:0.05001):0.00846):0.02796):0.02981):0.01676):0.03773):0.00171,Tarsius:0.13472):0.0070):0.00426);

end;

Without outgroup, you only describe the differences between the two taxa,. Only with outgroup can you describe the relationships between the two taxa. How you choose your outgroup decide how you can describe the ingroups relationships. For example, you make different conclusion about Human-Chimp relationships when you use monkey or fish as outgroup.

If you want to use phylogeny to help resolving your species limits questions you will need one, or more outgroups to root the tree. You may alternatively not use phylogeny, and use other species limits tests, depending on how many genes you are using (GYMC, PTP, multiloci tests); you may consider looking at sequence divergence, describing discrete and testing quantitative morphology, etc. Of course you can also code morphological characters and add to your phylogeny - but again, if you use a phylogenetic approach you will need an outgroup for that matter.