Global change could affect the structure and function of ecosystems in complex ways. In particular, gradual climate warming and fragmentation of ecosystem habitats are known to impact strongly on aquatic populations (Shuter and Post, 1990). To tackle this question, among numerous possible modeling techniques, we opted for a dedicated, stochastic, spatial, individual-based model of the global bullhead population dynamics at the river network scale (Grimm, 1999). The bullhead population we studied was that living in the River Bez network (France); it was resident and totally closed, and so immigration and emigration were not considered. We introduced temperature fluctuations and their impact on various demographics traits, such as growth, reproduction and survival. Several effect submodels were defined to describe these relationships, leading to a set of 40 parameters. The model was written in C++, and provided several population characteristics, such as individual numbers per age-class and network compartment over the course of time. The simulation results were compared to field data, and the model was validated by a complete sensitivity analysis based on a cluster analysis followed by a co-inertia analysis (Klepper, 1997; van Nes and Scheffer, 2005). Our individual-based spatial stochastic model provided a good description of variations in both population dynamics and age-class distribution among the patches by including temperature fluctuations. This model appeared to be robust to variations in the initial condition, and displayed the expected colonization process. The study of individual origins revealed a consistent dispersal pattern directly linked to the geography of the Bez network: downstream patches showed a high level of mixing between the individuals. Through sensitivity analyses we highlighted two main parameters; the temperature coefficient of the juvenile survival rate, and the temperature variability, which has the opposite effect on the population dynamics. Finally, the co-inertia analysis added the age as a variable of great interest for understanding the network age-class distribution.