Analyzing the functional type concept in arctic plants using a dynamic vegetation model


  • Howard E. Epstein,

  • F. Stuart Chapin III,

  • Marilyn D. Walker,

  • Anthony M. Starfield

H. E. Epstein, Dept of Environmental Sciences, Univ. of Virginia, Charlottesville, VA 22904-4123, USA ( – F. S. Chapin III, Inst. of Arctic Biology, Univ. of Alaska, Fairbanks, AK 99775-7000, USA. – M. D. Walker, Inst. of Northern Forestry Cooperative Research Unit, Univ. of Alaska, Fairbanks, AK 99775-6780, USA. – A. M. Starfield, Dept of Ecology, Evolution and Behavior, Univ. of Minnesota, St. Paul, MN 55108, USA.


Grouping organisms into categories based on common traits has been a tool of ecological scientists for some time now. Defining groups of species, examples being life forms and functional types, is an operational procedure, conducted to answer a particular question. This becomes rather practical when performing analyses at coarse spatial scales, given data limitations and the potential for species redundancy. However, the implications of aggregating organisms for modeling purposes are still unclear. Does averaging the traits of several species into a functional type sufficiently represent the dynamics of the individual components? How much variability is lost when we aggregate species into groups? In an attempt to address these questions, we examined how the level of vegetation aggregation affected a variety of ecosystem properties using a regional-scale model of arctic tundra ecosystems (ArcVeg).
We used four levels of aggregation: species (15 dominant ones), functional types (7), life forms (4) and vegetation type (1), in addition to two methods of aggregation (simple vs weighted means of plant parameter values). We found that the level of aggregation consistently affected community composition, total community biomass and ecosystem net primary production (NPP). Neither simple means nor weighted means of aggregated parameter values adequately captured the ecosystem properties simulated at lower levels of aggregation. Aggregation of vegetation (i.e. reduced parameter variability) using simple means underestimated total biomass, whereas aggregation using weighted means overestimated total biomass. Aggregation led to increases in NPP with both methods. These findings suggest that aggregating vegetation, particularly to levels less detailed than plant functional types, will have important implications for regional-scale modeling of vegetation dynamics and carbon cycling.