Crop wild relatives (CWR) are an indispensable and at the same time threatened genetic resources for plant breeding. The study uses wild species related to celery to demonstrate how genetic resources of CWRs can be actively maintained in their natural surroundings (in-situ). Genetic reserves should be designated for long term conservation of selected occurrences. The study presents the selection procedure in detail, aiming at the identification of occurrences and sites suitable for the designation of genetic reserves, the spatial model of a genetic reserve and first practical results of the project. The overall aim of the project is the establishment of a nationwide network of genetic reserves for Apium graveolens, Helosciadium repens, H. nodiflorum and H. inundatum, the four wild celery species native to Germany. Helosciadum repens (Jacq.) W.D.J.Koch is threatened by genetic erosion due to a decline in population numbers and sizes. The loss of any population is an irretrievable loss of diversity and opportunity to enhance crops in the future. Genetic reserves are one way to conserve these populations and their genetic potential. Twenty-seven populations were selected for the analysis in a decision process based on site information. Microsatellites (SSR) were used to elucidate the genetic diversity of German populations. A cluster analysis was performed to see if the individuals form clusters of similarity. For that, a discriminate analysis of principal components (DAPC) was conducted, as the inbreeding index indicated a high number of inbreeding events in the populations and thus discordance with HWE (Hardy-Weinberg equilibrium). The analysis identified six genetic groups, which coincide well with the geographic origin of the analysed plants. The allelic richness (mean counts of alleles per individual per population) was higher in the southern populations compared to the northern ones. This North-South discrepancy was also visible as a high heterogeneity in the cluster assignments in the DAPC analysis. These differences in genetic diversity might be a result of the biogeographic history of Europe, especially the last glacial maximum. For the establishment of genetic reserves, two populations were considered as most important: The population that differs the most from the average genetic composition and the population that represents the average genetic composition of a population the best. The two extremes of differentiation were interpreted as such that the former has a specific adaptation to its local environment, and the latter represents all populations the best. DifferInt was used to analyse the SSR data and validate the differentiation of all populations compared to a pool of populations. However, SSRs are not capable of detecting adaptive traits. Populations were additionally chosen from different eco-geographic units (EGU), to increase the chance of capturing different traits. EGUs (Naturräume) are areas of specific abiotic and biotic features. These features may influence selection pressures and induce local adaptations. Based on site parameters and genetic data, 14 most appropriate wild populations (MAWP) were identified for genetic reserves establishment. For H. repens, two eco-forms are known and described in the literature. Besides their different habitats (terrestrial/semi-terrestrial and aquatic) they can be differentiated by morphological traits. Leave and stolon sizes and flowering behaviour differ significantly. Furthermore, the roots of the aquatic forms do not anchor in soil but on other aquatic plants, wood or roots of trees, while the terrestrial form exhibits a shallow root system network similar to other perennial species. To this end, no genetic analysis was conducted to clarify the phylogenetic status of the putative forms and authors avoided the usage of any specific noun rather than form. The SSR data from the previous study was evaluated, particularly with regards to the two forms. Additionally, an ISSR analysis was conducted, and the data was used to perform a PCA. There was no genetic clustering regarding the two forms neither in the SSR nor in the ISSR data. Nonetheless, the North-South discrepancy in the genetic diversity that was visible in the DAPC plot was confirmed in the PCA of the ISSR data. However, markers may fail to detect quantitative variation for adaptively important traits. As the most obvious difference in the two habitats was the water availability, the adaptation of both forms to drought stress was studied by measuring the relative water content of leaves, system water content and water loss during drought stress conditions. The stomatal index was measured for different water treatment levels. The results indicate that phenotypic plasticity rather than genotypic adaptation is responsible for different H. repens phenotypes.