Since 2010, Cleveland Metroparks has been collecting vegetation monitoring data from a spatially-balanced random sample of 400 plots established throughout the 23,000-acre park system. This monitoring effort was designed to track community changes over time and to inform management decisions. This study represents an attempt to establish a standard classification and description of forest types in Cleveland Metroparks.
Aerial cover and woody stem data were collected from 1000 m2 plots, typically in a 20 meter by 50 meter layout. Relative cover and importance values were calculated and used in the multivariate analyses. Hierarchical clustering and indicator species analysis were used to identify and describe forest types. Nonmetric multidimensional scaling and random forests classification were then used to further explore the forest groups and identify possible environmental factors influencing plant community development. The descriptions were then cross-walked to alliances and associations within the U.S. National Vegetation Classification (USNVC) hierarchy.
Results/Conclusions
After the removal of non-forest plots, five distinct forest types were identified through hierarchical clustering and indicator species analysis. The composition of what we describe as alluvial forest is highly variable but quite distinct from most of the non-alluvial forest types and includes many exotic and ruderal species. Ordination and random forests classification also suggests that alluvial forests tend to have more open canopies and are associated with high pH soils that developed through alluvial processes. The four non-alluvial forest types showed less compositional variability and separated primarily along slope and soil moisture gradients. A comparison with established USNVC associations suggests that our forests have some ruderal elements that present classification challenges. Additional classification challenges arise because the sampled forests appear to be transitional between well-described associations of the Midwest and those of the Northeast. Furthermore, we recognize that due to the random nature of the plot placement, our sampling efforts are likely not capturing the best examples of the forest types in our park system which likely injects some ambiguity into our classification.
The results of our analyses allow us to establish a preliminary baseline forest classification for the park system that can be used to inform future monitoring and management efforts. We have also been able to identify information gaps and weaknesses in the current classification that will help to refine future data collection efforts. Finally, these data and results can be used to support modifications or improvements to the existing USNVC descriptions.