The results of phylogenetic analysis using morphological data are highly congruent with those based only on sequence data - read Brooks, D.R., J. Bileweitch, C. Condy, D.C. Evans, K.E. Folinsbee, J. Fröbisch, D. Halas, S., Hill, D.A. McLennan, M. Mattern, L.A. Tsuji, N. Wahlberg, J. Ward, D. Zamparo, and D. Zanatta. 2007. Quantitative phylogenetic analysis: progress and prognosis. Rev. Mex. Biodivers. 78: 225-252. Indeed, morphological data can be understood as correlated multi-gene complexes and sequence data can be understood as micro-morphology.

Thank you Mr. Hearty!! I have read about it, and I was thinking also that the more characters the best, cause as I´m not doing any molecular analyses I have to gather more information from the morfological.

I was reading right now your article from Santa Maria Island in Azores, and it conduced me to another in Canarias, and to the species Harpa doris and it´s morphological changes during the MIS 5.5... 

Thanks for your answer!!

In my view you can build a tree which in sense give a good idea of phylogeny of your group it would be definitely worthy and valid. it would be better if you go for Molecular phylogeny if it possible. More over the choice of characters you take don't show homoplasy. Just sort out good number of characters and then choose the characters that you think it is important. minimum number of character depent upon your perception of figuring out minimum of 15 or so for a matrix table. it it would be worthy if the little work has been done in your field.

The results of phylogenetic analysis using morphological data are highly congruent with those based only on sequence data - read Brooks, D.R., J. Bileweitch, C. Condy, D.C. Evans, K.E. Folinsbee, J. Fröbisch, D. Halas, S., Hill, D.A. McLennan, M. Mattern, L.A. Tsuji, N. Wahlberg, J. Ward, D. Zamparo, and D. Zanatta. 2007. Quantitative phylogenetic analysis: progress and prognosis. Rev. Mex. Biodivers. 78: 225-252. Indeed, morphological data can be understood as correlated multi-gene complexes and sequence data can be understood as micro-morphology.

Thank you Mr. Brooks and Mr. Asif!!! I will check for this article you suggest, and I´ll try to develop a good idea of all this. I have related 20 characters from the shell morphology and internal anatomy of some organs and the reproductive system. I think with it I could reach to suitable conclusions. Thank you all for the help!!!

Technically, all you need is three terminal taxa and two characters to infer a resolved phylogeny. So, on the face of it (20 species and 20 characters), you're in good shape. You might want to look at your matrix and see how the characters appear to be distributed, whether they are informative, or not. You could then reduce computation time by eliminating autapomorphies.

Of course you can use morphological characters to generate a phylogenetic hypothesis, but whether it is (1) fully resolved or (2) historically accurate are further questions.  Chances that 20 binary characters can fully resolve relationships of 20 ingroup taxa are very remote because homoplasy happens; many apparent convergences and reversals are due to variability (including discrete polymorphism) of morphological characters being retained across speciation events. I like ordered multistate characters where supported by logical, topological or quantitative relationships among states, but some people object to them and insist on binary data (either philosophical hardasses or the logic-challenged; opinions differ). Inappropriate choice of outgroups and methods of character coding can skew results, but at this point it is most important to use whatever information you have to generate a detailed hypothesis for subsequent testing; and if you get a chance to preserve tissue samples while doing the morphological work, the DNA studies can be done later when all your species have been wiped out by exotic predators or sea-level rise.

Thank you for you help... now that I am working on it, I have seen that some character varies very little between the different species, I have tried to choose multistate characters and I have gather information, from the shell, the sexual organs, and I have also included geographical information. I don´t have access to DNA materials cause it will means to make a  research in all the island of Cuba from east to west and sampling the types locations to collect DNA samples from the fresh snails. It will be a nice project but right now I don´t have the economy nor the time to carry it out... may be in the future...but I think a basic phylogenetic hypothesis can be done, also there are some studies were DNA materials had been collected but just in 5 species, I could set that as a baseline and build another approach and if in the future I get money to carry out this project I´m sure will be very nice!!! Thank you all for your help!!!!

Dear Jonathan,

Studying a genus with a phylogeny can be a good idea. In your case, since you are studying a genus which has not been reviewed in a long time. So, you'll probably should select a few outgroups for your analysis. This way, you'll be able to tell if your genus is monophyletic or not.

Here are a few advices from what I learnt and experienced when coding morphological matrices. There's more to write about it, but it is just the things that pop to mind right now.

For taxonomic meaningful result, I suggest you use type specimens at some point, including the type species of the genus - may sound self evident, yet it's important, and not always done it published research.

Be also cautious which characters you select : you should avoid sexual dimorphism (or clearly code the character for the male and female separately); you should also avoid to code twice the same character by selecting two characters completely linked... This last part can be tricky, for instance if a first character state the presence of a particular structure and the next code properties of this structure : in this case, when the structure is lacking, you'll have to code not applcable (generaly "-" in the matrix, but "?" is processed just the same) for the second character, not "0", because if you do that, you will just give more weight to the absence of the structure... In the same spirit, some people prefer to decompose multistate characters... Could be worth trying to see if your multistate character can be coded as a few binary characters... You may discover that you inadvertently mixed up two different things.

You won't be able to avoid homoplasy - that's not a problem, that just being honest.

After you obtain a parsimonious tree or trees, don't consider it to be the truth, it is only as good as your coding. So, I suggest you check how characters are mapped on the tree and whether this makes sense. You may end up correcting a few things. 20 characters x 20 taxa = 400 observations, it's easy to do a mistake!

Best of luck

Does group you are working on include living and fossil taxa or just living ones? My experience is with primates for which we have lots of molecular data including sequence, SiNE an LINE data. I have run comparisons between large datasets >200 morphological characters and 16 taxa. The results are not very encouraging. The well established molecular trees are mostly but not completely in agreement with the morphological. I can see many examples of convergence in feeding systems for example. See Molecular Phylogenetics and Evolution January 2015, Vol.82:358–374, doi:10.1016/j.ympev.2013.12.002

Phylogeny and Biogeography of Neotropical Primates.

The CIs are fairly low and do not fall in a consistent way with respect to anatomical parts. I always doubt that a high CI means anything other than that the author was very selective about the characters and coding, often meaning that they had a preconceived notion about the phylogeny based on long study of the group.

so, for fossil analyses, I use a molecular scaffold that takes into account that we know so much more about the living taxa than we ever will about extinct ones, but should never ignore the fossils., even incomplete ones. Uncertainty is to be expected. An example of my take on this (again for primates, sorry) is in the March 6th Perspective in Science Magazine. Please check out the supplemental docs. I posted the whole thing on ResearchGate already. 

There ar many ways to do this, but that's my take.

Yes, while I agree with (and appreciate) Dr. Brooks' perspective, in my limited experience your phylogeny really depends on the group and the available characters! It won't hurt to try building a character matrix in MacClade and analyzing with maximum parsimony in PAUP* (I know there are many others...). I don't know how publishable a morphological phylogeny will be in this day and age, unless there is some independent hypothesized evolution of the group out there (e.g., from molecular data). 

As far as how many characters, I agree with Dr. Kay that you need to use all available characters regardless of your preconceptions. The more the better.

Thank you all for your help!!! I will bear in mind all your comments... just one more thing, we are talking about landsnails in Cuba, that phylogenetic studies had been poorly done, mainly because DNA test are very expensive for Cuban researchers   and by the other hand there aren´t plenty of mollusks specialist there... I just want to make an approach that could help understanding this group. Thanks again!!!!!

Morphological characters are useful and important for defining taxa, especially at genus level. The ideas behind molecular phylogeny are same as those for morphological classification. As long as you can identify conserved morphological traits that are unique for this group compared with their most closely related species, you have the genus (or at any other level) well defined.