Acral melanoma (AM) is an extremely aggressive subtype of melanoma that is prevalent in east Asia. AM exhibits high intertumoral and intratumoral heterogeneities with poor prognosis. To associate the genomic heterogeneities with phenotypic traits and efficacy of treatments, a method is needed to recover genomic information from limited samples with high specificity and sensitivity from early stage AM specimens. We performed laser capture microdissection (LCM) to isolate single micro-tumor-nests, containing only dozens of cells, from stained tissue slices and then applied multiple annealing and looping based amplification cycles (MALBAC), a highly efficient whole genome amplification methods originally developed for single cells, to amplify the whole genome of each tumor nest for sequencing. We were able to accurately profile the landscape of copy number alterations (CNAs) and single nucleotide variations (SNVs) of every single micro-tumor-nest, and to quantitatively characterize the heterogeneities at different levels, between tumor and nevi, between patients, between different phenotypes within a same tumor, and between adjacent tumor cell clusters with identical phenotypic appearance. We have found that genomicheterogeneity exists extensively and branched evolution happens in the early-stage of AM development. We are able to build the phylogenetic tree among these phenotypically addressable cell clusters