It depends on your paper topic. If you just want to describe that there is a new gene family then not necessarily, but if you want to talk about the evolution of your new gene family then definitely yes.

If you have an outgroup, you can root the tree with that outgroup, and only then you can interpret your tree as a phylogeny (without an outgroup it is just a dendrogram, not a phylogeny). To discuss evolution using a tree, you need a direction to decide which nodes are truly ancestral and which are not. This is only possible to accomplish with an outgroup.

I didn't quite understand the data basis of your (one?) tree. Did I understand correctly that you have data for four species and all copies of your gene family for those four species? What do you mean with "another genes"? Do you have data for other genes for your four species in the same alignment/tree?

It depends on your paper topic. If you just want to describe that there is a new gene family then not necessarily, but if you want to talk about the evolution of your new gene family then definitely yes.

If you have an outgroup, you can root the tree with that outgroup, and only then you can interpret your tree as a phylogeny (without an outgroup it is just a dendrogram, not a phylogeny). To discuss evolution using a tree, you need a direction to decide which nodes are truly ancestral and which are not. This is only possible to accomplish with an outgroup.

I didn't quite understand the data basis of your (one?) tree. Did I understand correctly that you have data for four species and all copies of your gene family for those four species? What do you mean with "another genes"? Do you have data for other genes for your four species in the same alignment/tree?

Thank you Norbert for your nice explanation. You understand correctly ( e.g I have used ALDH genes from four species gene family G.raimondii (My own species), T.cacao, Arabidopsis and Rice). And another genes mean other three species gene family. Yes I have data for other genes in same tree.

I think, two things are mixed here together that can't be mixed. What you have is a gene similarity tree, not a phylogeny. A phylogeny is a tree that is derived from an alignment that compares only homologous sequences (so it basically applies only to single copy genes). A phylogeny aims to show relationship of taxa, however, you want to show the relationship of genes. What you need to show is whether your gene similarity tree with several copies from four gene families is congruent to the (true) phylogeny.

That your tree splits into four branches according to the four gene families (not your four taxa) is obvious because the gene families lack any similarity (maybe except for some motifs). Your approach is therefore technically not wrong but the 4-split into the genes is obvious. I don't know whether this is a standard in phylogenomics, but I don't think its wrong either.

Additionally to your gene family tree(s), you need a phylogeny. For your four taxa, you don't need to calculate a phylogeny, just refer to APG III (10.1111/j.1095-8339.2009.00996.x), which is more or less the current standard for the grand scale plant phylogeny. Rice is a monocot, which is sister to the eudicots, to which your other three species belong. Gossypium and Theobroma both belong to the Malvaceae, so they should be sister to each other, Arabidopsis is a Brassicaceae, which is a bit further away and therefore sister to the group Gossypium+Theobroma.

(Rice) (Arabidopsis(Theobroma+Gossypium))

The interesting part is now how your gene copies are arranged in your gene similarity tree in comparison to the true phylogeny. Now that you can compare the similarity on your genes with the true phylogeny, you can interpret the evolution (gene family member duplications/losses etc.) of your new gene family.