• White, Alan S.
  Department of Forest Ecosystem Science - University of Maine, Orono, Maine
  
• Witham, Jack W.
  Department of Wildlife Ecology - University of Maine, Orono, Maine
  
• Hunter, Jr., Malcolm L.
  Department of Wildlife Ecology - University of Maine, Orono, Maine
  
• Kimball, Alan J.
  Department of Forest Management, University of Maine, Orono, Maine
  

Several researchers have hypothesized that plant species richness has a unimodal relationship with biomass, while others have argued for a linear relationship. Data from various types of herbaceous communities show some supports for the unimodal hypothesis, but this has not been tested extensively for forests and questions remain concerning its generality. We used linear and quadratic regression models to examine the relationship between overstory biomass and richness in a coastal Maine Quercus-Pinus forest across and within cover types using a data from two plot sizes (2500-m. sq. quadrats and 625-m. sq. sub quadrats). Understory data from 1-m. sq. plots were analyzed. Richness was quadratically related to biomass at both plot sizes for all cover types combined, but the amount of variation explained by the models was very low (R sq. < 0.09) Richness and biomass were not significantly related at either plot size for the mixed mesic cover type, the most common type in the forest. The best fit (R sq. = 0.43) was obtained with a quadratic model for the conifer cover type at the sub-quadrat level, with the quadratic model for the 1-m. sq. data having the second highest R sq.(0.24). Across all six data sets, the quadratic model was the only one with a significant fit in two cases and had considerably higher R sq.'s (1.3 - 1.9 x) in two others. The remaining two data sets could not be fit with a significant model of either type. For this forest, these results suggest little support for a linear relationship between plant species richness and biomass and variable, often weak, support for a unimodal relationship. Density, a potentially confounding variable in this type of analysis, was only weakly correlated with richness and was not found to alter the relationship between biomass and richness.