Communities are influenced by many factors, with anthropogenic impacts being one of the strongest. These factors can influence community structure and may cause non-random species loss by filtering certain traits of species, which may also determine how a species contributes to ecosystem functioning. The structure of a community can also be highly variable over short timescales and across seasons, as abiotic factors can alter a resource over the course of days and also alter intensity of competition within a community. These short-timescale influences are most noticeable in an ephemeral resource that is limited in both space and time. Changes in community structure and interactions can be represented as networks of interactions (links) among species (nodes). Interaction networks describe a community and incorporate non-trophic interactions, which can alter their structure. These are generally measured by counting the number of trophic interactions, ignoring non-trophic interactions such as competition. However, competitive interactions may be important for network dynamics, yet the most appropriate way to quantify competition remains unclear. The outcome of a competitive interaction could potentially be predicted by the body size ofcompeting individuals, and this would remove the need to observe individual interactions. These ideas were tested using the dung-associated community in an Afromontane forest reserve in Nigeria across variation in seasons and in areas that were protected from anthropogenic impacts by fencing. Trapping and recording of interactions within the community was used to assess community structure, and experiments were run to test how dung removal and secondary seed dispersal changed across seasons and in protected areas of the forest. The influence of competition was determined by altering the size and number of dung beetles present at the resource. The community structure was influenced by forest protection and varied across seasons. Additionally, the size and number of dung beetles present was higher in protected areas in the wet season, resulting in higher dung removal and secondary seed dispersal and indicating non-random species loss in unprotected areas. The attractiveness of the dung resource changed over short timescales and desiccation occurred rapidly in the dry season, which resulted in decreased insect abundance and diversity over the course of days. Individual dung removal rates were not additive, and could only be used to predict community-level dung removal when taking into account competitive interactions. The body-size ratio between winners and losers of a competition was a good predictor of an interaction outcome. Competition networks were highly connected and nested, with compartmentalisation occurring in the competition network. At the community level, effects of forest protection and changing seasons only altered network nestedness. Therefore, body size can be used to predict species responses to anthropogenic threats and community structure and function when taking into account density-dependent competitive interactions. Furthermore, when determining community responses to anthropogenic threats, sampling across changes in seasons and observing competitive interactions provides more information about a community structure and stability.