Species and plant community distribution in a Mojave-Great Basin desert transition

Multivariate analyses were used to describe the vegetation characteristics of a transition from lowelevation Mojave desert to higher-elevation Great Basin desert. Vegetation data used were from Plutonium Valley in the Nevada Test Site. Data from forty nine releves were analyzed with two classifications (two-day indicator analysis or TWINSPAN and unweighted paried group cluster analysis or CLUSTER). Three ordinations, reciprocal averaging (RA), detrended reciprocal averaging (DCA) and non-metric multidimensional scaling (MNDS), were also used. A rotational correlation analysis was used to determine the vector direction of environmental gradients that correlate best with ordination results. Only token correspondence was found between multivariate classes generated by TWINSPAN and CLUSTER, and seven classes (plant communities) identified from field reconnaissance. The latter seven communities were based on differences in dominant species. Distribution of the vegetation was related more to beta diversity than alpha diversity. Individual species were much less diagnostic than the amount of plant cover, groups or guilds of species or differences in elevation and steepness of slope. Because of the high beta diversity the NMDS ordination gave results with the greatest ease of interpretation.