Skógræktarritið - 15.05.2001, Page 148

tional range. New influences like
tourism, cabin „villages", forestry,
and an increasingly finely woven
web of roads have also emerged.
In addition to these direct inter-
actions between man and birch
forests, expected future climatic
shifts towards generally milder
winters, regionally increased
level of summer precipitation
and a higher frequency of ex-
treme weather events also repre-
sent a potential influence on
birch forests even at the ecosys-
tem level and at a continental
scale (cf. Skre 2000). Thus, on
this background of stronger and
potentially more severe influence
on mountain birch forests from
man, it is of great importance to
develop scenarios for future sus-
tainability of various manage-
ment regimes.
The model will take into ac-
count main factors influencing
forest productivity, and various
direct and indirect human inter-
actions with the birch forest.
These interactions include an-
thropogenic direct and indirect
factors like domestic reindeer
and sheep herbivivory and tram-
pling, forestry, tourism and other
vegetative influences. Interac-
tions between ungulate and
insect herbivory, and periodically
strong impacts from outbreak
species like the autumnal moth
Epirrila autumnata (e.g., Tenow el
al. 2000, Neuvonen et al. 2000)
will also be included in the
model. The model will also be
applied to simulate scenarios for
a changing climatic regime due
to global warming, including its
direct and indirect effects on
birch forest productivity, distrib-
ution and abundance, and pat-
tern of herbivory.
Model perspectives
Models in general contribute to
the objectivity of a theory. The
mountain birch forest model
assessment against data provid-
ed by the 20 project participants
and the literature provides a test
of the model's effectiveness.
Three levels of assessment can
be made for complete models
(Ford 2000): fitting, predicting,
and revealing different results.
These three topics will be de-
scribed below using scaling
problems and complex popula-
tion dynamics as an illustrative
example.
Fitting is not a strong assess-
ment criterion for a specific
ecosystem theory. Yet it can be
difficult to achieve and when it is
achieved there has to be a thor-
ough understanding of how that
was done. Fitting is more like an
alternative mathematical and
computational description of a
given verbally formulated model
describing a system with its sug-
gested intrinsic functional rela-
tionships.
Even if fitting is considered
being a weak assessment criteri-
on, it will be an important aspect
of the HIBECO mountain birch
ecosystem model. The model will
not be a realistic model in the
sense that fitting is meant to
reproduce a specific mountain
birch forest system in a specific
area to as great detail as possi-
ble. Rather, it will be a model
that is able to simulate what will
be considered the most impor-
tant elements shaping the forest
system today and in the future in
a "representative" virtual land-
scape and its socio-economic
and cultural context. Thus, fitting
in this case refer to being able to
simulate the system's key pro-
cesses in general terms, where a
delicate balance between realistic
model details and generalizing
power of functional principles for
this ecosystem is maintained.
When formulating the model
one is forced to be explicit about
which components (forcing and
state variables) to include in the
model and which to exclude.
Further, one is forced to be ex-
plicit about formulation of the
system's functional relationships
(flowcharting). Parameters' and
state variables’ spatial and tem-
poral variability in statistical
terms must be documented from
real data or "educated guesses",
and compared with model simu-
lation outputs in the validation
phase.
Prediction is more valuable
than fitting and is widely used in
both statistical modeling and
system simulation as vafidation.
The HIBECO birch forest model
wili be of the latter kind (spatio-
temporal computer simulation).
Hopefully, it will contribute to
shed light on hypotheses related
to complex relationships in this
ecosystem, including scale-relat-
ed problems.
For example, when validated
and verified against historic time
series and environmental condi-
tions for local insect outbreaks,
can one be reasonably confident
that it will be able to predict the
next outbreak in a specific area,
given the necessary parameter
adjustments and other necessary
background data? The autumnal
moth outbreaks happen with a
periodicity of 9-10 years at
regional to local scale in parts of
Fennoscandia (Neuvonen et al.
1999, 2000 and references there-
in), while the outbreak intervals
are more complex at the even
finer scale of birch forest stands
(e.g., Tenow and Bylund 1989,
Tenow et al. 2000) and at very
coarse scales (Neuvonen et al.
1999). Various ways of formulat-
ing the local birch/moth/para-
sitoid/climate interactions in the
model may contribute to verify,
falsify or modify hypotheses
related to proposed synchroniza-
tion factor(s) and reasons for
outbreaks under various local
146
SKÓGRÆKTARRITIÐ 2001 l.tbl.