Agriculture Reference
In-Depth Information
is 2.0 years. Graphically, this is equivalent to the total area of the annual histograms
divided by the initial leaf number, which is essentially the same as the method
shown in Figs.
3.2
and
3.3
.
Such long-running observation series intended for a dynamic life table analysis
sometimes are stopped for practical reasons when half the leaf cohort has died
(Kohyama 1980; Diemer 1998a,b); truncating the observations precludes calcula-
tion of the age-dependent probabilities of leaf death, but the observed leaf half-life
provides a useful estimate of leaf longevity in its own right (Diemer 1998a; Dungan
et al. 2003). On the other hand, the dynamic life table approach applies equally well
to short series of observations over days, weeks, or months rather than years. Miyaji
and Tagawa (1973, 1979) constructed dynamic life tables for leaves of
Tilia japon-
ica
and
Phaseolus vulgaris
, both species with short-lived leaves. The longer obser-
vations continue, the more risk that dynamic life table analyses will be confounded
by stochastic variation in the risk of mortality across the years of observation.
Dynamic life table analyses are not only confounded by stochastic variation but
also biased by differential rates of leaf mortality in better versus worse leaf
microenvironments (Takenaka 2003). Thus even if leaves are selected randomly to
establish the sampled cohort, the sample will concentrate into “better” places over
time. Static life table analyses are not immune to the problem of stochastic interan-
nual variation, but they do not suffer this sampling bias.
The data required for static life table analyses are gathered in one round of
sampling, which makes this approach logistically appealing. Static life table
analyses do not follow a single leaf cohort over its lifetime but instead reconstruct
the life table from different aged cohorts of leaves observed at a point in time.
Unfortunately, the record of growth cycles in tropical regions usually is too obscure
or ambiguous to apply the static life table approach with confidence. In tropical
forests, the number of leaves on a branch whorl does give information about leaf
emergence pattern, but the seasonal timing of leaf emergence is not fixed in species
or even on branches in a single tree (Kikuzawa et al. 1998). For example, in
Araucaria araucana
the mean interval between successive whorls was not exactly
1 year, and varied among individual trees depending on their light regime (Lusk and
Le-Quesne 2000). On the other hand, the required sampling is relatively easy to
apply with evergreen trees in boreal and temperate regions where the basic approach
has a long history of use (Pease 1917). In these strongly seasonal climates, clearly
visible terminal bud scars typically demarcate annual growth increments along the
shoot (Fig.
3.4
); it is easy to reconstruct the ages of growth segments along a
branch, and hence the age of the leaves on each segment. We then can infer the
number of leaves in each annual cohort by counting the number of leaves still
attached and the number of leaf scars left by fallen leaves in each shoot growth
increment. This static approach, however, assumes no year-to-year variation in leaf
demographic parameters, which can be problematic because of interannual climatic
variation, age-dependent loss of leaves to herbivory, or trade-offs in resource
allocation between production and reproduction. Kayama et al. (2002), for example,
found this assumption did not hold for some evergreen conifers. Interannual
variation can also confound leaf longevity estimates from a dynamic life table
