A priori, we identified species groups among eight halophytes of the southe
rn California tidal marsh plain, using cluster analysis on growth form and
life history traits which are commonly employed to produce 'functional grou
ps'. We then tested the efficacy of the groupings by assessing biomass and
nitrogen accumulation Functions of each species in two greenhouse experimen
ts that differed in soil moisture and nutrient availability. Plants were gr
own for 6 mo in both monotypic and multispecies microcosms. Functional attr
ibutes differed among species, including members of the a priori groups. We
found a 2-fold range in tissue nitrogen concentrations and a 2.5-fold rang
e in biomass accumulation. Rhizosphere root distributions differed, and roo
t-to-shoot ratios ranged 3-fold. The presence of neigh boring species alter
ed the biomass or nitrogen concentrations of four species. A posteriori clu
ster analysis based on microcosm performance produced different species gro
ups than found in the a priori analysis. Two species pairs appeared functio
nally redundant, although the annual and perennial pair would likely differ
entiate in time, and species in the second pair tend to be found at differe
nt elevations across the marsh plain. We argue that the designation of func
tional groups should be based on measured performance and objective classif
ication procedures. Rather than assuming functional redundancy, we recommen
d testing for similarity of group members under Varied conditions, e.g., al
one and with their common neighbors and under benign and stressful conditio
ns.