Identifying behavioural basis of competitive relationship is essential to understand outcome of interspecific competition. However, it remains difficult to investigate demographic effect of competitive behaviour, because various kinds of behaviours may co-occur in the competition and make the dynamics far complicated in nonlinear ways. We report that the behavioural basis of interspecific interaction can be identified, by focusing on the timescale difference from the occurrence of each behaviour to the appearance of its demographic effect. Between two bean beetles, Callosobruchus chinensis and C. maculatus, major interspecific interactions are resource competition (RC) at the larval stage and reproductive interference (RI) at the adult stage. RC has longer time lag than RI, because effect of RC appears in the adult number of the next generation through larval competition while effect of RI appears instantaneously in the adult number through early death of females. If we detect two effects with different time lags from the competition dynamics, an effect with intergenerational time lag and with no time lag would be considered as RC and RI, respectively. We applied empirical dynamic modelling approach, which is a nonlinear time series analysis for detecting causal interactions and the strength, to two published datasets of experimental competition between those beetles. Results showed the significant causality from the winner species to the loser one in both experiments, but the causality time lag differed between experiments: the causality had no time lag in the C. chinensis-win data, while intergenerational time lag in C. maculatus-win data. Furthermore, detection of the causality with intergenerational time lag from C. maculatus to C. chinensis in both experiments suggests interplay of constant RC and variable RI which can reverse the outcome. This study is the first successful case study that links behavioural-level interactions to demographic-level effects in interspecific competition.