Eusociality (Greek eu: "good" + "social") is a term used for the highest level of social organization in a hierarchical classification. The term "eusocial" was introduced in 1966 by Suzanne Batra and given a more definitive meaning by E. O. Wilson. It was originally defined to include those organisms (originally, only invertebrates) that had certain features: 
The most familiar examples are insects such as ants, bees, and wasps (order Hymenoptera), as well as termites (order Isoptera), all with reproductive queens and more or less sterile workers and/or soldiers. The only mammalian examples are the naked mole rat and the Damaraland mole rat.
The phenomenon of reproductive specialization is found in various organisms. It generally involves the production of sterile members of the species, which carry out specialized tasks, effectively caring for the reproductive members. It can manifest in the appearance of individuals within a group whose behavior (and sometimes anatomy) is modified for group defense, including self-sacrifice ("altruism").
Subsequent to Wilson's original definition, other authors have sought to expand or narrow the definition of eusociality, focusing on the nature and degree of the division of labor, which was not originally specified. A narrower definition specifies the requirement for irreversibly distinct behavioral groups or castes (with respect to sterility and/or other features), and such a definition excludes all social vertebrates (including mole rats), none of which have irreversible castes. A broader definition allows for any temporary division of labor or non-random distribution of reproductive success to constitute eusociality, and some have accordingly argued that even humans may be considered eusocial. Others believe that the hierarchical classification may not serve much purpose.
In spite of the obvious advantages of common foraging and defense, eusocial animals had appeared paradoxical even to Darwin: if adaptive evolution unfolds by differential survival of individuals, how can individuals incapable of passing on their genes possibly evolve and persist? Since they do not breed, their fitness should be zero and any genes causing this condition should be eliminated from the population immediately. In Origin of Species (first edition, Ch. 8), Darwin called this behavior the "one special difficulty, which at first appeared to me insuperable, and actually fatal to my theory." Darwin anticipated that a possible resolution to the paradox might lie in the close family relationship, but specific theories (e.g. kin selection or inclusive fitness) had to wait for the discovery of the mechanisms for genetic inheritance.
Early ideas on eusociality included suggestions that trophallaxis or food sharing was a basis for sociality. Other theories include superorganism theory and parental manipulation theory. The most widely accepted model to explain eusociality is based on W.D. Hamilton's idea of inclusive fitness.
According to inclusive fitness theory, eusociality may be easier for species like ants to evolve, due to their haplodiploidy, which facilitates the operation of kin selection. Sisters are more related to each other than to their offspring. This mechanism of sex determination gives rise to what W. D. Hamilton first termed "supersisters" who share 75 per cent of their genes on average. Sterile workers are more closely related to their supersisters than to any offspring they might have, if they were to breed themselves. From the "selfish gene's" point-of-view, it is advantageous to raise more sisters. Even though workers often do not reproduce, they are potentially passing on more of their genes by caring for sisters than they would by having their own offspring (each of which would only have 50% of their genes). This unusual situation where females may have greater fitness when they help rear siblings rather than producing offspring is often invoked to explain the multiple independent evolutions of eusociality (occurring some 11 separate times) within the haplodiploid group Hymenoptera — ants, bees and wasps. However, Hymenoptera is a large group and the majority of hymenopterans are not social. Furthermore, highly developed eusociality also exists in non-hymenopterans, perhaps most prominently in termites. Certain vertebrates (such as the naked mole rat) have also been described as eusocial. Most such cases involve organisms that display high levels of inbreeding, such that colony members share more than 50% of their genes, and therefore the same model is considered to apply to these species.
Reeve and Holldobler's version of superorganism theory further elaborates this model by considering competition and co-operation between groups as well as within groups.  In this case, an individual's inclusive fitness varies depending on how much it invests in within-group competition (e.g. hoarding a private food cache) versus between-group competition (e.g. contributing to common foraging); and on its relatedness to the other group members. In a hymenopteran colony with one breeder (queen) and many workers as described above, the evolutionarily stable state is for each individual to invest entirely in helping the group, leading to a perfect "superorganism", which implies the stability of eusociality in this case. This agrees with Hamilton's model. This is implied even without considering between-group interactions. However, they further show that any group of relatives may show high "superorganismness", provided that there are many groups competing for the same resources. This may favour eusociality, or a degree of eusociality in non-hymenopterans. Indeed, a non-zero level of inter-group co-operation is predicted, even if the group members are entirely unrelated, as long as there is competition between groups.
Theories of parental manipulation point out that the transition from solitary to eusocial appears to involve intermediate stages where dominance interactions are required to suppress the reproductive tendencies of group members; that is, females are manipulated into acting as workers, even if it is against their own self-interest.  This model does not require that individuals be highly related, though high relatedness will reduce expected levels of resistance to manipulation.
Some hypotheses about how eusociality evolved in naked mole rats include: inbreeding, ecological factors such as the dependence on large tubers that are hard to locate and reach underground, heat loss prevention, and high dispersal costs. In the mammalian cases, eusociality is believed to arise from 'reproductive suppression', where infertility in working females is only temporary, and not genetic.
Recently, some species of gall-making aphids (Order Hemiptera) and thrips (Order Thysanoptera) were found to be eusocial, with many separate origins of the state. These species have extremely high relatedness among individuals due to their partially asexual mode of reproduction (sterile soldier castes being of the same clone as the reproducing female), but the gall-inhabiting behavior gives these species a defensible resource that sets them apart from related species with similar genetics. In these groups, therefore, high relatedness alone does not lead to the evolution of social behavior, but requires that groups occur in a restricted, shared area.
Similarly, eusociality has arisen among some crustaceans and other arthropods. On some tropical reefs, several species of minute Synalpheus pistol shrimp that depend on certain sponges for the survival of their colony, live eusocially, with a single breeding female and a preponderance of male defenders, armed with enlarged snapping claws. Again, there is a single shared domicile for the colony members, and the non-breeding members act to defend it.