Meeting report
Tree-grass interactions workshop,  15-19 November 1999, NCEAS, Santa Barbara


The working group analyzing tree-grass interactions in savannas and savanna-like systems, NCEAS,  November 1999.
Standing, L to R: Richard Joffre (CEFE-CNRS, Montpellier), Bob Scholes (CSIR, South Africa), Mike Cougenhour (CSU), Jose San Jose (IVIC, Venezuela), Bill Parton (CSU), Jo House (Kings College, London), Jaques Gignoux (Ecole Normale Suprieure, Paris), Joe Scanlan (Dept of Natural Resources, Queensland), Guillaume Simioni (Ecole Normale Superieure).
Sitting, L to R: Jonathon Scurlock (ORNL), Chad McMurtry (Texas A&M), John Ludwig (CSIRO, Darwin), Ruben Montes (Universidad Simon Bolivar, Caracas), Steve Archer (Texas A&M), David Breshears (Los Alamos).
Not present in photograph: Nancy Kiang (UCB).

It was the final formal meeting of a series with started with a SCOPE-sponsored planning session in Paris, in 1996, a NCEAS workshop in May 1998, and a modeling workshop in South Africa in November 1998. The completion of this synthesis will take place through electronic ‘virtual meetings’ through mid-2000.

One of the originators of the group, Prof David Hall of Kings College, London, passed away in August. In recognition of his contribution, one of the key outputs of the working group will be dedicated to him.

The meeting concentrated on two main activities: a meta-analysis of the large savanna site dataset which has been accumulated through the activity; and detailed theory-testing using four savanna models applied at two sites.

The savanna dataset, which is unique, will be housed at the ORNL-DAAC on completion of the project. Considerable effort has been put into ensuring that it is well-organised and self-describing. It has three levels of detail. At level 1, it consists of about 140 records describing the climate, soil and vegetation of globally-distributed savanna sites. This datatset has been used to define the environmental envelope in which tree-grass systems exist, and predict a variety of their structural features. A byproduct is a set of stand-level allometric relationships between tree basal area, leaf area, canopy cover and biomass.

Levels 2 and 3 consist of increasingly detailed experimental and observed production data from key savanna sites (about 6 locations, and a total of about 20 sites). It is used to support the model analysis, which has focused on how the models capture the characteristic resource-partitioning between woody and non-woody plants in savannas, and the consequences this has for predictions of total system NPP. The experimental data from the arid site at Boatman, Queensland, where five levels of tree clearing were imposed and monitored over a 10-year period, and the moist site at Lamto, Ivory Coast, where patches of naturally-different tree density have been monitored over several decades form the basis of a robust model intercomparison.

Apart from the database and the improved models, the key products of the workshop will be papers: