Context <p>Pollinator populations are generally sensitive to landscape fragmentation, but the contribution of habitat types at different area proportion to the development of bumblebee colonies at landscape scale is unclear.</p> Objectives <p>We contribute to this field of landscape ecological research by analysing how the area proportions of semi-natural habitats, flower strips and mass-flowering crops affect the density of bumblebee colonies in artificial landscapes with a simulation model.</p> Methods <p>We created different artificial landscapes (1&#xa0;km × 1&#xa0;km) with gradients of the area proportions of the different habitat types in a fully crossed design (landscape composition). Additionally, we created a gradient of habitat fragmentation by modifying the number of fields per landscape and, thus, the number of habitat patches (landscape configuration). Then, we applied the open source model BumbleBEEHAVE to simulate the population development of <i>Bombus terrestris</i> over five years in the artificial landscapes.</p> Results <p>Each habitat type showed a significant positive effect on the simulated number of bumblebee colonies with increasing area proportion in the landscapes. In contrast, the effect of the number of habitat patches (habitat fragmentation) was negative. The efficiency of flower strips was higher in landscapes with low area proportions of semi-natural habitats and oilseed rape. Both food (nectar) and nesting habitat were limiting resources for the density of bumblebee colonies in the landscapes.</p> Conclusions <p>Our results show that additional food and nesting resources provided by flower strips are particularly important for bumblebees in landscapes with low proportions of semi-natural habitats and oilseed rape as mass-flowering crop. To substantially increase the number of bumblebee colonies, not only more food but also more nesting habitat is required. This result is in line with previous results from real landscapes with and without biodiversity measures.</p>