The abundance of a given species is usually expressed in terms of either th
e number of individuals per unit area or volume (i.e., density), or biomass
. These two abundance metrics generate different results at both the statis
tical analysis level (e.g., comparison of means) and the ecological level (
e.g., diversity comparisons). we seek here to unify different abundance met
rics using the formula A = N(B/N)(k/3) where A is the abundance of a given
species, k represents a fractional dimension, N is the number of individual
s per unit of area and B is the biomass of the sampled species. When k = 0,
A is density and when k = 3, A is the biomass. A value of k = 1 would give
abundance approximately proportional to the sum of the length of individua
ls and k = 2 would give abundance approximately proportional to the sum of
their surface ureas. Metrics intermediate between density, length, area and
biomass are possible using non-integer values of k. Applying this methodol
ogy to ichthyological data characterized by highly variable intraspecies bi
omass, we examined the effect of the abundance metric on the results of a t
hree-factor analysis of variance (depth, season and site). In some cases, d
ifferences which could not be seen with either density or biomass could be
seen with intermediate metrics. we suggest that many ecological results cou
ld be usefully evaluated in terms of the effect of the fractional dimension
of sampling. In some cases, such an approach could identify the optimal me
tric.