Most studies of density-dependent demography in plants consider the density
only of the single focal species being studied. However, density-dependent
regulation in plants may frequently occur at the level of the entire commu
nity, rather than only within particular species. In addition, because dens
ity dependence may differ considerably seven in direction) among demographi
c parameters, generalizing about patterns of density dependence and extrapo
lating to lifetime fitness and to population dynamics require comparisons a
mong life history stages, as well as among types of species and physical en
vironments.
We constructed seminatural communities of desert annuals composed of all th
e constituent species in the same relative proportions as found in the natu
ral habitat. These experimental communities were planted at a range of dens
ities that extended far above and below mean natural field density. We comp
ared among physical environments (irrigation treatments), among communities
from different physical environments, and among growth forms (dicot and gr
aminoid) to search for generalizations about the magnitude and direction of
density dependence.
Strong evidence of community-level density dependence was detected at all t
hree life history stages studied in these desert annuals: emergence, surviv
al, and final size, However, both the direction and degree of consistency o
f this density dependence varied considerably among the stages. The stronge
st and most consistent competitive effects were experienced at the emergenc
e stage, where the mechanism is most likely a form of interference competit
ion rather than exploitation competition. At the survival stage, the magnit
ude of effects was highly variable among physical environments and source c
ommunities but negative effects were relatively rare, with either positive
or no significant effects of increasing density. Thus, exploitation competi
tion was also unimportant at the survival stage. In contrast, for growth, e
xploitation competition appeared to be the primary mechanism of interaction
influencing growth. This variation in mechanism, direction, and magnitude
of interactions among life history stages suggests that current models of p
lant community structure that are based largely on exploitation competition
as it influences growth (with mortality a simple function of growth) are i
nadequate for even this simple annual plant community.
We also compared growth forms and found that graminoids were superior compe
titors to dicots at the emergence and survival stages; they also had higher
emergence and survival, regardless of density. Consistent with this result
, grasses are always the numerical dominants in the source communities. In
contrast, the two growth forms did not differ in competitive ability for gr
owth, and dicots were consistently larger individuals, independent of densi
ty, even though grasses were also usually the biomass dominants in the sour
ce communities. These results suggest the importance of nontrophic mechanis
ms of interaction in controlling community structure and again emphasize th
e importance of constructing and testing models that incorporate multiple m
echanisms of interactions.