communities (two compositions at two diversity levels) and an environmental
warming treatment (constant temperature or warmed). Five replicates of each
treatment combination made a total of 40 microcosms. Population densities
were sampled in the second and sixth week of the experiment.
Experiment 4
This experiment tested the insurance hypothesis: that biodiversity stabilizes
aggregate community and ecosystem properties (Petcheyet al., 2002). There
were four diversity levels (two, four, six, and eight species) and four composi-
tions at each level. This is the only experiment that contained a single trophic
level of protist species (bacterivores). In addition, there were three environ-
mental variability treatments (constant temperature, fast temperature fluctua-
tions and slow temperature fluctuations). Population densities and CO 2 flux
were estimated in the second and sixth week of the experiment.
Population consequences of body size
We analyzed the relationship between cell mass and population size within and
across the four experiments using three separate analysis of covariance models.
In all analyses, data points were averages of the values of a species across the
replicates of each treatment combination in which it occurred. That is, there
is only one data point per species per treatment combination. All analyses
included log 10 abundance as the response variable, and log 10 body size as a
continuous explanatory variable. The first model included three data sets
(Experiments 1, 2, 3) but only constant temperature regimes. The explanatory
variables were trophic group (categorical, five levels, autotroph, bacterivores,
herbivores, omnivores, predators), experiment (categorical, three levels, 1, 2, 3)
and species richness (continuous). The second model included only the data set
from Experiment 3 and explanatory variables were environment (categorical,
two levels, constant temperature and higher temperature) and species richness
(continuous). The third model included only the data set from Experiment 4 and
explanatory variables of environment (categorical, three levels, constant tem-
perature, fast varying temperature, slow varying temperature) and species rich-
ness (continuous). Significant interaction terms between body mass and another
explanatory variable would indicate that variables affect the population con-
sequences of body size. All interaction terms were initially included and back-
wards elimination by Akaike Information Criterion (AIC) and F-test were used to
find the minimum adequate model (Crawley, 2002 ).
Community consequences of body size
Food webs that receive similar amounts of energy should reach similar levels in
total biomass, regardless of differences in organism size found among different
food webs (Damuth, 1981 , 1987; Brown & Gillooly, 2003 ; Brown, 2004 ; Mulder
CONSEQUENCES OF BODY SIZE IN MODEL MICROBIAL ECOSYSTEMS 249