164 Kaoru Kitajima and Lourens Poorter
vary amon gspecies in a continuous manner
(Augspurger 1984a). As a result, the abun-
dance and establishment probabilities of seedlings
exhibit different, yet overlapping, distributions
amon gspecies in relation to li ght availability
(Montgomery and Chazdon 2002, Poorter and
Arets 2003). In other words, the competitive
edge of one species over another at any par-
ticular light availability appears to be a mat-
ter of probability, which may be very subtle.
What types of trade-offs lead to such continu-
ous variations in preferred light environment of
seedlings?
TRADE-OFFS PROMOTING
SPECIES RICHNESS WITHIN A
HORIZONTAL PLANE
Species-specific traits associated with size and
biomass allocation patterns are thought to under-
lie various trade-offs that contribute to species
sortin galon ga niche axis. Niche theory posits
that coexistence of species A and B is possible
when species A outperforms species B in one envi-
ronment, but species B outperforms species A in
a second environment. Individual fitness compo-
nents, such as growth rates, survival rates, or
fecundity of individuals, can be used to evaluate
performance of potentially competin gspecies in
contrastin genvironments. However, these indi-
vidual performance measures may not be pos-
itively correlated with each other, nor equally
important in their relative contribution to over-
all fitness at different positions alon ga niche
axis. Thus, two types of trade-off must be distin-
guished.
In the first type of trade-off, adaptations to
one type of environment preclude optimal trait
combinations in another environment, leadin gto
a rank reversal in a fitness component between
the two environments (Latham 1992, McPeek
1996, Chesson 2000). The second type of trade-
off occurs between two fitness components, such
as growth rates and survival (Brokaw 1987,
Kitajima 1994, Poorter and Bongers 2006). If the
relative importance of these fitness components
shifts between two environments, it can lead to
an overall performance rank reversal. A hybrid of
these two types of trade-off is often reported in the
literature as a stron gempirical pattern in relation
to light environment, that is, high light growth
versus low light survival trade-off (Kobeet al.
1995, Wright 2002, Baralotoet al.2005, Hubbell
2005).Butdoesthistrade-off existbecausespecies
that grow fast in high light are somehow pre-
vented from growing fast in the shade (the first
type of trade-off), or because growth and survival,
two fitness components, exhibit negative cross-
species correlation regardless of the environment
(the second type)? These two alternative hypothe-
ses predict contrastin gcross-species correlations
when growth or survival is compared between two
contrastin genvironments (Box 10.1).
For quantitative evaluations of these two alter-
native hypotheses, how survival and growth
respond to light gradients must be quantified
for multiple species, as shown in Figure 10.2.
These types of response curves are known in pop-
ulation ecology as phenotypic reaction norms,
and a rank reversal (i.e., crossin gof reaction
norms) represents a stron gcase of genotype×
environment interaction (Schlichtin g1986).
Many species increase growth rates as light avail-
ability increases from deep shade (0.5–2% full
sun) up to the light levels found in treefall gaps
(e.g., 10–40% full sun), followed by a plateau
and possibly a decline when light levels exceed
the optimum (Figure 10.2b). Survival also tends
to increase at higher light availability associated
with larger treefall gaps, but shows no response
or even a decline with gap size in some species
(Figure 10.2a). In general, species differences in
growth rates tend to be larger at higher light avail-
ability, while species differences in survival tend to
be greater at low light than at high light.
How stron gis the evidence for rank reversals
of growth or survival rate along a light gradient
when these two key performance traits are exam-
ined separately? For objective evaluation of the
frequency of rank reversals, either parametric or
non-parametric statistics may be used (Box 10.1).
All statistical tests require that a sufficient number
of species are compared between two contrasting
environments that represent the two purported
niches (e.g., gap versus shaded understory), or
when possible, alon gthe entire li ght gradient
observed in the community.