eLearning

INTRODUCTION
Most ecological experiments involve isolation and
manipulation of a small part of an ecosystem, for
example in bottles or mesocosms in the case of
aquatic ecosystems. Results are then directly extrapo-
lated to whole ecosystems [in the sense of Likens
(1992)] or even larger systems. My experience with
aquatic experiments at several scales is that such
extrapolation is often questionable, for important
features of whole lakes and their communities are
usuallymissing. Although this is partly a function of
size, microcosms and mesocosms also typically lack
the complexity of whole ecosystems, so that such
features as air–water and sediment–water ex-
changes and the activities of wide-ranging organ-
isms are not included. Others have reached similar
conclusions about the representativeness of meso-
cosms (Frost and others 1988; Benndorf 1990;
Carpenter 1996; Englund 1997; Peterson and others
1997; Lodge and others forthcoming; Pace forthcom-
ing). Indeed, as I shall discuss, many of the results
from experiments in small whole lakes need to be
corrected for differences in hydrodynamics, gas
exchange, and other processes in order to be prop-
erly extrapolated to larger lakes (Fee and Hecky
1992; Fee and others 1996).
In several previous articles, I have compared
results from long-term experiments done in small
Received 26March 1998; accepted 14May 1998.
*e-mail: d.schindler@ualberta.ca
Ecosystems (1998) 1: 323–334
ECOSYSTEMS
r 1998 Springer-Verlag
323whole lakes with results from mesocosms or micro-
cosms (Schindler 1988, 1990; Levine and Schindler
1992). However, discussions with participants at the
recent ecosystem experiment workshop (Jasper Na-
tional Park, Alberta, 14–17 November 1997), sev-
eral recent papers on the subject (Carpenter and
others 1998; Lodge and others forthcoming; Pace
forthcoming), extensions of my experience to com-
munity studies in mountain lakes (Paul and Schin-
dler 1994; McNaught and others unpublished),
recent studies of scaling over a range of lake sizes
(Fee and Hecky 1992; Fee and others 1996), and
reviewer comments that I read as subject matter
editor for articles in this volume have motivated me
to update my views. I believe that the pivotal role of
spatial and temporal scales in ecosystem experi-
ments is underappreciated, yet critical to the ad-
vancement of ecology. In particular, whole-ecosys-
tem experiments appear to be losing favor because
they often cannot be exactly replicated and are
expensive and difficult to execute, leading many
ecologists to favor smaller scales in order to obtain
the satisfaction of statistical confidence.