In mutualistic networks of pollination and seed dispersal, recent studies showed that highly nested structure was amazingly ubiquitous. To examine evolutionary causes and consequences of the nestedness, we developed a simple mathematical model using adjacent matrices. The model showed that connecting directly or indirectly with the main part of the community is necessary for a species to survive in the community. This finding contradicts previous ideas that the nested networks were more robust than random ones. If the stability is not the evolutionary driving force of the nested structure, what is the cause? One of the possible explanations is that the observed nestedness is an artifact due to difficulty to observe interactions between rare species. The other explanation is that there is some advantages for specialists to interact with generalists rather than with specialists. Generalist populations are supported by many partner species and contribution of some specialists is not important, while specialist species are supported by a few species and contribution of each partner species is essential to maintain their populations. In other words, some species are more parasitic than others in the mutualistic community. Future studies may reveal dominance of parasitism in superficially harmonious mutualistic communities.