Biogeography of the vascular flora on islands in the Bering Sea Region

The floristic data for 29 islands and six source areas in the Bering Sea Region were compiled primarily from the published works of Hulten. Information from recent collections in the Aleutian Islands was also added to the floristic data set.

Geographic variables included area, maximum elevation, distance to source areas and other islands, and latitude. Measures of topographic complexity were calculated from samples of the number of contour lines per unit area on topographic maps. Indices of the size and complexity of island shape were developed from radial measurements of the islands. The size and shape of an island were also represented by the number of bays and total length of mapped streams. The depth and width of interisland passes, percentage of near shore area, distance along current paths and distance to the limit of pack ice were used to represent the local isolation of the Aleutian Islands.

The relationships between the floristic and geographic variables were analyzed with correlation and multiple regression. Patterns in the floral similarity of the islands were examined with cluster analysis.

The biogeographic patterns of the vascular flora of the islands in the Bering Sea Region are similar to the ones predicted by the theory of island biogeography that was developed by MacArthur and Wilson. The number of species is positively correlated with the area. The log-log model is the best form of this relationship. The number of species is also positively correlated with maximum elevation and other indices of environmental heterogeneity. Correlations between the total number of species and the distance to source areas are not significant. Significant negative correlations are observed between the proportion of species related to a source area and the distance from the source area. Similar to the prediction of the theory of island biogeography, "near" islands are observed to have a significantly steeper species-area curve than "far" islands.

The results of multiple regression analyses indicate the following order of the importance of the types of geographic variables in the prediction of the number of species: area, complexity of island shape, distance to source areas, and maximum elevation or topographic diversity. The width of the interisland pass is the most important variable representing the local isolation of an island in the multiple regression models.

The results of cluster analyses show that neighboring areas tend to have greater floristic similarities than more distant areas. These results suggest the existence of a pattern of isolation across the region. Differences in the isolation of groups of species with different dispersal abilities are observed from the results of cluster analyses. The results of cluster analyses also indicate that the flora of the Aleutian Islands tends to be more similar to the North American source areas than the Asian source areas.

Significant positive correlations between the number of species in families and the total number of species are observed. The significance of correlations among the numbers of species in families is not known or previously reported.

Geographic variables related to the size, environmental heterogeneity, and isolation of an island can be used to predict the number of species on islands in the Bering Sea Region. The relationships between the floristic and geographic variables are also similar to the ones used by MacArthur and Wilson in the development of the theory of island biogeography.