Results/Conclusions The results clearly supported the concept of bundling roads, i.e., several roads bundled close together, or an upgraded road with more traffic on it. Therefore, large un-fragmented parts of the landscape should be kept un-fragmented. The results also showed that the thresholds strongly depend on the behaviour of the animals at roads, e.g., the degree of road avoidance, which, in turn, depends on road characteristics such as traffic volume. However, landscape metrics do not capture this relationship and therefore, their predictive value is limited. In addition, the effects of road networks have long response times leading to an extinction debt. Transportation authorities are taking advantage of this lack of knowledge when they build new roads, arguing that not enough is known about the thresholds, and more research would be needed before they would slow down. This constitutes a “fragmentation spiral”, because research is (and may always be) unable to catch up. This situation is contrary to the precautionary principle. Therefore, the German Federal Environment Agency has recently suggested to introduce region-specific limits to control landscape fragmentation and urban sprawl. At the same time, empirical studies comparing landscapes with differing road network configurations should be conducted to test the model predictions, to improve the model, and to provide a better foundation for planning highway networks. This presentation makes a strong case for introducing limits to curtail landscape fragmentation and urban sprawl for the time until the extinction debt and the thresholds are known.