To identify and characterize the plant communities on California hardwood rangelands, and to determine if those communities are random sets of species or distinguishable vegetation associations, we sampled 455 100-m transects in the Coast Range and Sierra Nevada Foothills during 2004 and 2005. Sampling points were established at 5-m interval along the transects. The number of times the vertical canopy projection of each woody species was intercepted by the transect was used as a measure of woody species composition. Herbaceous species composition was estimated using the species-rank method. Of the 455 transects 267 included woody plants. Of a total of 54 woody and 254 understory species only 14 woody and 60 understory species were present in at least 3% of the transects. These species were used in a vegetation matrix to perform the analyses. Ordination was performed using non-metric multidimensional scaling (NMDS). Manhattan distance was chosen to obtain the dissimilarity matrix from the matrix of species abundance. A Monte Carlo test using a randomized version of our data set was used as a null model. The NMDS for the matrix of woody and understory species extracted stronger axes (lower stress values) than the one for the randomized matrix, indicating that the assemblage of species is not random. Observation scores on the same axes were used to perform a cluster analysis to group the transects in vegetation associations.

Results/Conclusions

The 267 transects with woody and understory species were grouped in 15 clusters based on the species composition. The clusters were very diverse and represent 11 oak subseries: 1) Blue Oak/Grass; 2) Blue Oak-Interior Live Oak/Grass; 3) Blue Oak-Foothill Pine/Manzanita/Grass; 4) Blue Oak-Valley Oak/Grass; 5) Interior Live Oak-Blue Oak-Foothill Pine; 6) Black Oak/Green leaf Manzanita; 7) Black Oak-Valley Oak/Grass; 8) Valley Oak-Coast Live Oak/Grass; 9) Coast Live Oak/Grass; 10) Mixed Oak/Grass and 11) Mixed Oak-California Buckeye/Grass. When analyzing the spatial distribution of the woody species, we noticed that Q. douglasii is homogeneously distributed in both, Coast Range and Sierra Nevada Foothills. The same is true for Q. wislizeni and P. sabiniana. Q. agrifolia is only present in the Coast Range while Q. lobata and Q. kelloggii, although present in both areas are mostly located in the Coast Range. Q. garryana is concentrated in the Northern part of the Coast Range. We conclude that NMDS is a method that can yield objective ordination and constitutes a good basis for the classification of hardwood rangelands.