Environmental Engineering Reference
In-Depth Information
tainty completely.This uncertainty has some important ethical implications
for decision-making. Let me illustrate with an example.
Suppose that a developer wished to convert a large tract of forested land
into housing and commercial enterprises and engaged a scientific consult-
ant to offer insights about the degree of fragmentation that could be sus-
tained without disrupting the diversity of animal species within the land
base. Suppose that the consultant executed an experiment like the one con-
ducted in Amazonia (chapter 6) to test the null hypothesis that habitat frag-
mentation did not cause loss in species diversity. Let's suppose that unlike
the Amazonian case, experimental fragmentation did not cause a change
in species diversity.We would then conclude that the evidence is insufficient
to reject the null hypothesis and thus permit development to go ahead.
But how reliable is the conclusion from this single test? The answer is:
limited.The reason is that we are unsure if the outcome is normal for this
kind of ecosystem or a rare event that depended on specific place and time.
Thus, a single experimental test of a hypothesis cannot provide a highly re-
liable answer.The only way to increase reliability of knowledge acquired
from ecological experimentation is to repeat the experiment many times.
That is why ecologists, when asked to offer scientific advice, often qualify
their conclusions with the disclaimer “this needs further study ....”,mean-
ing, this needs more replication to gain general insight into the likely trend.
Such a disclaimer is a perennial
source of frustration for policy mak-
ers who seek definitive answers to
specific problems.Yet the disclaimer is
legitimate because ecologists can
never be definite that one factor or
variable is the causal driver of a pat-
tern or process. This is because re-
peated experimental tests of the same
hypothesis will never produce identi-
cal (invariant) outcomes in the field due to random environmental effects.
Consequently, ecologists are only able to present their scientific results in
terms of the likelihood or risk that a particular event will happen (Ludwig
et al. 2001). Indeed, this is true for any scientific discipline that must deal
with confounding effects of random environmental factors encountered in
natural environments.
A single experimental test of a
hypothesis cannot provide a highly
reliable answer. The only way to
increase reliability of knowledge
acquired from ecological experi-
mentation is to repeat the experi-
ment many times.
