The relative importance of different processes in shaping metacommunity structure may strongly depend on spatial scale. We analyzed the spatial architecture of metacommunities of cladocerans in high-altitude temporary peat land pools of the Tunari mountain range (Bolivia). The spatial distribution of pools in this part of the Andes is hierarchical: pools are clustered within small peat lands. These peat lands lay scattered within valleys that are separated from each other by mountain ridges. In our study, we analyzed two datasets (the PEATLAND- and VALLEY-dataset), each with a different spatially hierarchic sampling design. For the PEATLAND-dataset, we exhaustively sampled all pools in each of 6 peat lands in four different valleys. For the VALLEY-dataset, we sampled single pools of numerous peat lands within the same valleys. The PEATLAND-dataset thus represents variation among communities within peat lands, whereas the VALLEY-dataset should contain the community variation that is present at the scale of valleys.

Results/Conclusions

When applied to the entire dataset, mantel tests indicated a relation between community dissimilarity and geographic distance among pools, even after correction for variation in important environmental variables. This relation was exclusively due to differences among valleys and proved to be invalid within valleys or within peat lands, suggesting that dispersal limitation may reduce the exchange of species among valleys, but not within valleys. Additive diversity partitioning revealed a disproportional high contribution of within-peat land diversity to gamma diversity (66%). The high within-peat land diversity was due to a higher than expected beta diversity among pools, concurring with a lower than expected alpha diversity. Average community dissimilarity correlated positively with an estimate of environmental heterogeneity within peat lands (r = 0.64). Given the absence of any spatial structure, these patterns suggest species sorting as main structuring force for cladoceran metacommunities within peat lands, rather than dispersal limitation or mass effects. Individual peat lands contained on average a similar amount of environmental heterogeneity than valleys. Concurrently, average community dissimilarity within peat lands was similar to average community dissimilarity within valleys. As a result, the unique contribution to gamma diversity of beta diversity among peat lands within valleys was low (14%) compared to the contribution of beta diversity within peat lands (47%).Our results indicate that conservation efforts should be mainly directed at the protection and maintenance of pool heterogeneity within peat lands. Furthermore, conservation of peat lands of different valleys should be preferred to conservation of different peat lands within single valleys.