Multisource Field Plot Data for Studies of Vegetation Alliances: Northwestern USA

Although vegetation alliances defined and described in the U.S. National Vegetation Classification are used as a fundamental unit of habitat for modeling species distributions and for conservation assessments, little is known about their ecological characteristics, either generally or individually. A major barrier to understanding alliances better is the lack of primary biotic and physical data about them. In particular, few alliance or association descriptions of the USNVC are based directly on original field plot data. Such data do not exist in the quantity or over the geographic extents necessary, and new field work to acquire such data is unlikely. This study attempts to learn about the efficacy of and limitations to developing the data needed by integrating existing information from multiple sources and themes across the Inland Northwest of the USA. Almost 40,000 field plot records from 11 different sources were integrated, sorted, and evaluated to generate a single standardized database from which plots were classified a priori as members of alliances. Additional data sets of climate, biomass productivity, and morphological traits of plant species were also integrated with the field plot data. The plot records were filtered for eight univariate parameters, species names were standardized, and multivariate outliers of species composition were identified and removed. Field plot records were extracted from the data sets with SQL statements based on existing descriptions of alliances, and these subsets were tested against a null model. Ultimately 21% of the field plots were classified to 49 vegetation alliances. Field plot classifications were corroborated with a nonmetric multidimensional scaling ordination. This study resulted in a large set of primary data critical for the study of vegetation alliances. It shows that it is possible to develop synthetic vegetation field plot data sets from existing multisource information.