Through its ability to share various resources, P2P networks have been in the focus of research for the last decades. Especially the utilization of search algorithms to retrieve the distributed resources in unstructured P2P networks is of major importance, since no global view of the network exists. This thesis addresses the need to evaluate and compare search algorithms for unstructured P2P networks with each other by using standard metrics. It provides two frameworks for systematic benchmarking and comparison of search algorithms in unstructured P2P networks. The first framework is implemented based on the Actor Model and the second one based on the Peer Model (a coordination based programming model). Both frameworks support easy exchangeability of search algorithms. The second goal of this thesis is to create a fully distributed search algorithm for unstructured P2P networks based on the collective feeding of bark beetles. Additionally, an already existing algorithm from a different domain based on the Physarum Polycephalum slime mold is adapted to fully distributed search for unstructured P2P networks. To achieve these goals the following methodological approach is applied. After an extensive literature research, the frameworks are first designed and afterwards implemented. As the next step, the two new search algorithms are developed and implemented into the frameworks. As the final step, both algorithms are benchmarked, evaluated and compared to four existing search algorithms: Gnutella Flooding, k-Walker, AntNet for P2P and SMP2P. Overall, Bark Beetle and the adapted Physarum Polycephalum algorithm show very good scalability regarding growing network size and load. In terms of absolute time, both algorithms show very promising results with only k-Walker having slightly better results for networks with high replication. In terms of success rate, Bark Beetle shows an almost equal good success rate as Gnutella Flooding, without having the same major drawbacks. Although the success rate for the Physarum Polycephalum adaption is very low for networks with small replication, it is significantly better for networks with high replication.