Why parents in some societies induce some of their sons to become religious celibates is an evolutionary puzzle. Some have speculated that this might be associated with brother competition for family resources. However, the behavioral ecology of monks and the possible links with competition between brothers remains unexplored. Here, we use demographic data from Amdo Tibetan agropastoralists in western China to evaluate what factors determine the probability of becoming a monk and explore the possible association between wealth and having a monk brother. We found that boys with at least one older brother are more likely to be induced by their parents to become celibate monks. Patrilocal heads of household, who inherit parental property, are more likely to be first-born sons, whereas men who marry uxorilocally, that is they move to their wife’s household, are generally second or later born sons. Moreover, we find that men with at least one monk brother are wealthier than men who only have non-celibate brothers. Together, these results suggest that sending a son to the monastery is way for parents to decrease competition between brothers over family resources. Harsh and resource-limited environments, like the one we consider, can lead to the emergence of communal households, including polyandrous families, which used to be common in Tibetan areas. Directing one son to become a religious celibate offers a potentially effective solution to brother competition in our population.

Socio-demographic data were collected in 2017 in a county in Gansu province, China. The study was approved by the School of Life Sciences, Lanzhou University, and the Research Ethics Committee of UCL (0449/003). We collected detailed sociodemographic data from 530 households in 21 natural villages in a county in Gansu province. Natural villages are clusters of houses that do not necessarily correspond to the larger administrative villages recognised by the local government. The households reported on a total of 3591 living people (1702 women and 1889 men). In each household, one adult man or woman was interviewed and asked about the age, sex, marital status, socioeconomic status and profession of all household members, including whether any members in the household were monks or nuns. In some cases, we were not able to obtain sibling information, because some people were absent at the time of the interview and the main interviewee could not recall this information. Interviews were conducted with the help of local translators, one adult man and one adult woman. Female interviewees were generally interviewed by a woman assistant. Participants were briefed regarding data anonymisation in the local language. Participants were briefed about the anonymity of our methods and data in local languages before giving consent. Informed consent was obtained from all participants.

Analyses reported in this article can be reproduced using this data. Notice that we reduced the number of covariates for the analyses in Tables 1-2 and have not included data for the analysis in Tables 3-4 to protect the identity of the participants.