The proliferation of mega-constellations comprising thousands of satellites has amplified the need for efficient routing solutions in space networks to minimize latency and ensure robust global connectivity. This work evaluates the performance of multiple decentralized routing algorithms within Walker-Delta constellations, comparing their effectiveness in Low Earth Orbit (LEO) and Very Low Earth Orbit (VLEO) configurations. Our findings reveal that constellation geometry, such as the number of satellites and orbital planes, shapes routing strategies more than orbital altitude, with VLEO demanding larger fleets to maintain seamless coverage. Furthermore, we propose PolarDisCo, a novel routing algorithm specifically designed to provide coverage of polar regions - a geographical area where, due to unique topology challenges, existing decentralized routing algorithms (e.g., DisCoRoute) exhibit a significant performance drop. Experimental results demonstrate that PolarDisCo reduces end-to-end delay on the polar area by 7.9 ms with respect to DisCoRoute, thus providing a 14.8% improvement. These findings underscore the potential of tailored routing strategies to optimize communication in next-generation satellite networks with LEO and VLEO shells.