AbstractSocial interactions, including agonistic behavior, are very important for the management and welfare of individuals forming groups in captivity. One of the main concerns for the stability and durability of adult male groups is a noticeable level of intraspecific aggression. This study comprises a Social Network Analysis to illustrate social structure in different groups of captive Saharan Dorcas gazelles (Gazella dorcas neglecta). Our main objectives were to examine the relationship between agonistic, affiliative, and association networks and their reciprocity, assessing also whether the agonistic networks can conform to a linear dominance hierarchy. For these purposes, we recorded the behavior of 23 adult males organized in five herds, three composed only of Saharan Dorcas gazelle males and two mixed herds in which there were also Mhorr gazelle males (Nanger dama mohor). Observations were carried out during 295 h through scan sampling. We found no correlation between the affiliative and association networks in any group, although there was a significant correlation between the agonistic and association networks in mixed‐species groups which was not present in single‐species groups. Overall, there was no consistent reciprocity in either affiliative nor agonistic networks and none of the agonistic networks showed a linear structure. These results indicate that affiliative behavior in Saharan Dorcas gazelles offers distinctive and valuable information about the bonds between individuals, however, their dominance structure is far more complex than previously thought. As information provided by affiliative and proximity behaviors is different in this species, we suggest considering affiliative interactions to stablish affinity between individuals. Evaluating different social behaviors and not only agonistic interactions in later studies, is also recommended to develop a more accurately daily management in zoos that guarantee group stability and individuals' welfare, which will improve the conservation of captive populations.