Multivariate techniques were used to investigate how environmental and disturbance factors affect the distribution and species composition of grassland, shrubland, and woodland plant assemblages at the landscape level.  Thirteen plant assemblages were defined and modeled across the landscape based on species composition from 425 stands.  The use of classification, ordination, and indicator species analysis produced relatively discrete and ecologically representative plant assemblages. Although variables related to moisture and soils were most important to plant distribution, vegetation patterns appear to be determined by a complex combination of edaphic, topographic, and climatic variables in addition to site history. Differences between assemblages provide a benchmark for better understanding variation between grassland, shrubland and woodland vegetation types in space and time.

Eric Pfaff and Paul Hosten
Southern Oregon University and Bureau of Land Management, Medford District, respectively

Grassland, shrubland, and woodland vegetation types are a major component of the landscape in the Applegate Valley of southwest Oregon.  The Ashland Resource Area of the Medford Bureau of Land Management is currently implementing thousands of hectares of fuels reduction treatments annually. However, there is a paucity of data regarding these vegetation types and how they are affected by fuel reduction treatments.  The objectives of this study were to gain a better understanding of these vegetation types and the factors responsible for their distribution.

Right: Study Area Map (larger view). Study area shown in red within the Klamath Mountains of southwest Oregon. The study area lies within the Applegate Valley of southwestern Oregon on U.S. Department of Interior (USDI) Bureau of Land Management (BLM) lands in the Ashland Resource Area.

The first author collected data on species composition, woody plant structure (size classes and form), and site characteristics for 425 non-conifer stands using a reconnaissance or releve style methodology. Thirteen plant assemblages based on species composition were classified using clustering techniques in conjunction with Indicator Species Analysis.  NMS ordination was used to investigate patterns of species distribution within assemblages. Nineteen environmental and ten disturbance variables were regressed against ordination axes and used to model assemblages based on landscape characteristics. PC-ORD software was used in analysis.

The figure below is an ordination based on the species composition of the 425 surveyed stands. Symbols represent stands, and different symbols represent different assemblages. The farther the distance between symbols, the more dissimilar the stands in terms of species composition. Pictures are from a representative stand of each assemblage (hover the mouse over the picture to see the assemblage name).  

Below: Summary of Plant Assemblages. Images of each of these assemblages follow. Also see the complete plant species list compiled for all stands.

Number of significant indicator species for each assemblage from indicator species analysis.

Average P values for each cluster analysis group from indicator species analysis.

Cluster Analysis, coupled with Indicator Species Analysis (ISA), clustered stands into thirteen groups (plant assemblages); this grouping had both a high number of significant indicator species and low average P values. This indicates a high degree of fidelity of indicator species to stand clusters relative to other numbers of groups.

Below: Mean percent cover by native/non-native functional group in the herbaceous layer for each assemblage. Periwinkle = Native Perennial Grass, Maroon = Exotic Annual Grass, Light Yellow = Native Perennial Forb, Light Blue = Native Annual Forb, and Dark Purple = Exotic Annual Forb. Assemblage labels are, from left to right: Fescue grassland, Buckbrush chaparral, Rabbitbrush/buckwheat bald, Manzanita/white oak, Oak woodland, Oat grass grassland, Mountain mahogany/Brewer's oak, Open oak/chaparral, Black oak woodland, Disturbance mediated woodland/chaparral, Madrone woodland, Manzanita chaparral, and Canyon live oak woodland.

Below: Mean percentage of oak size classes for each assemblage Sizes are diameter at breast height (DBH). Light Blue = DBH Greater than 16", Light Yellow = DBH 9-16", Maroon = DBH 4-9", Periwinkle = DBH Less than 1". Assemblage labels are, from left to right: Fescue grassland, Buckbrush chaparral, Rabbitbrush/buckwheat bald, Manzanita/white oak, Oak woodland, Oat grass grassland, Mountain mahogany/Brewer's oak, Open oak/chaparral, Black oak woodland, Disturbance mediated woodland/chaparral, Madrone woodland, Manzanita chaparral, and Canyon live oak woodland.

• Variables related to moisture (elevation, summer evapotranspiration, and annual precipitation) and soils (pH, soil horizon depth, and total silt and clay) were found to be the most important determinants of non-coniferous vegetation distribution.

• Disturbance (wildfires and fuel reduction treatments) appears
  responsible for increasing the variability of species
  composition within assemblages.

• The high proportion of exotic annual grasses in the
  disturbance mediated woodland/chaparral assemblage
  suggests that disturbance may increase the proportion of
  exotic taxa.

• Non-native taxa are mostly in the form of exotic annual
  grasses found typically in low elevation, low canopy situations.

• Regression results indicate that heatload (aspect and slope)
  and fire history are associated, yet neither are highly
  correlated to plant distribution.

• Classification and Indicator Species Analysis produced
  ecologically representative assemblages, and environmental
  variables elucidated factors predictive for patterns of species
  distribution within assemblages. An exception to the
  classification working well is the fescue grassland
  assemblage, which had a high elevation, mostly native
  component and a low elevation, mostly naturalized
  (non-native) component.

• Predictive habitat models of assemblages based on ordination scores produced mixed results.  This is possibly due to assemblages occupying multiple ecological sites, individual species responding independently to predictor variables, inaccuracies in base data, limited site history information, unmeasured biotic processes, and variation within stands and associations. 

• Oak stands are present as mixed size classes. Typically oaks belong to the small size classes with fewer large trees.

• The rarity, high species richness of native plants, and the potential for the extirpation of the fescue grassland and black oak woodland plant assemblages warrant efforts to ensure their persistence on the landscape.

Link to E. Pfaff thesis (PDF).
Contact E. Pfaff or P. Hosten.

Photos were taken by E. Pfaff, unless otherwise noted.

Dr. Pat Muir and Dr. Paul Hosten, Principal Investigators