Developing a data-driven classification of South Florida plant communities

Sah, Jay; Ross, Michael; Stoffella, Susana

A comprehensive, broadly accepted vegetation classification is important for ecosystem management, particularly for planning and monitoring. South Florida vegetation classification systems that are currently in use were largely arrived at subjectively and intuitively with the involvement of experienced botanical observers and ecologists, but with little support in terms of quantitative field data. The need to develop a field data-driven classification of South Florida vegetation that builds on the ecological organization has been recognized by the National Park Service and vegetation practitioners in the region. The present work, funded by the National Park Service Inventory and Monitoring Program - South Florida/Caribbean Network (SFCN), covers the first stage of a larger project whose goal is to apply extant vegetation data to test, and revise as necessary, an existing, widely used classification (Rutchey et al. 2006). The objectives of the first phase of the project were (1) to identify useful existing datasets, (2) to collect these data and compile them into a geodatabase, (3) to conduct an initial classification analysis of marsh sites, and (4) to design a strategy for augmenting existing information from poorly represented landscapes in order to develop a more comprehensive south Florida classification. Thirty five data sets, comprising vegetation data from 8,246 sites, were received from researchers working in various organizations. The structure and completeness of available data sets were examined in terms of sampling design, the number and size of sampling units, spatial distribution, taxonomic resolution, and method of estimating species abundance. The data were then summarized at the site level and were incorporated into a geo-database. Finally, vegetation classification for marsh sites was developed using a hierarchical cluster analysis of the sites-by-species matrix. The analytical summary of the datasets in geodatabase provided the basis for identifying current gaps in the vegetation data needed to develop a comprehensive south Florida vegetation classification. The results from the cluster analysis of species data for marsh sites were used to cross-walk with the recently updated and hierarchical Vegetation Classification System for South Florida Natural Areas (Rutchey et al, 2006), to evaluate whether vegetation classes at various levels identified in this system were well substantiated by the classification achieved through field-data based cluster analysis. While the existing vegetation data covers most of terrestrial natural areas and vegetation associations present in South Florida, some areas are more extensively represented than others and some vegetation types, in both woody and herbaceous groups, described in existing classification systems are missing in the datasets. Moreover, classification of herbaceous sites using agglomerative cluster analysis suggests that freshwater and salt marsh communities are arranged along hydrology and salinity gradients, respectively. In addition, plant communities in some localities are largely reflections of different types and levels of disturbance. To be most useful within the ecosystem restoration efforts currently underway in the region, a comprehensive classification system should incorporate those elements, so that tangible changes in community composition due to alterations in the environmental drivers can be easily assessed.

Developing a data-driven classification of South Florida plant communities

Collections