eLearning

Over the past 31 years, I have conducted experi-
ments with aquatic ecosystems at several spatial and
temporal scales, including short incubations in small
bottles, in situ mesocosms of several sizes, and
whole-ecosystem experiments of several types [re-
viewed by Schindler (1988)]. Although each of the
scales has proved important in our attempts to
understand in-lake processes and catchment–lake
interactions, each has shortcomings. This became
apparent as we used mesocosms to forecast and
fine-tune whole-lake experiments. Conversely, we
have used whole lakes to calibrate and verify that
smaller-scale experiments were properly represent-
ing the interplay of ecosystem-scale processes to an
extent where results could be extrapolated with
confidence to ecosystem-scale issues [for example,
see Schindler (1988, 1991) and Levine and Schindler
(1992)]. I believe that such calibration is one of the
most important roles for whole-lake experiments,
for in most ecosystems, experiments must be re-
stricted to small scales. For reasons that are difficult
to anticipate, smaller-scale experiments often yield
erroneous conclusions about community and ecosys-
tem processes. Others, working in other aquatic
ecosystems, have drawn similar conclusions (Car-
penter 1996; Lodge and others forthcoming; Pace
forthcoming). In some cases, elaborate measures
have been devised to make mesocosms simulate
even rather simple ecosystem-scale processes accu-
rately, for example in the Marine Ecosystem Re-
search Laboratory (MERL) estuarine facility (Nixon
and others 1984; Oviatt 1994; Santschi 1985). As
Pace (forthcoming) has emphasized, knowing the
spatial and temporal scales to which an experimen-
tal result can be extrapolated with confidence is
very important to ecology.
The following discussion is based largely on re-
sults at the Experimental Lakes Area (ELA), which
was one of the first facilities to be devoted primarily
to ecosystem-scale experiments. The reason for the
facility was a very simple one: those responsible for
managing large and important water bodies were
reluctant to risk large sums of money on manage-
ment schemes that had been tested only in short-
term bottle or mesocosm experiments. Small whole
lakes were seen as a necessary intermediate step
(Johnson and Vallentyne 1971). But, because lakes
and resources were limited, we often used smaller
experiments as pilot studies for whole-lake experi-
ments, providing the basis for the following discus-
sion.
Importantly, I believe that managers and politi-
cians in today’s conservative, economically ori-
ented, resource-hungry society are even harder to
convince to act in an ecologically responsible man-
ner than those of 30 years ago, making it more
necessary than ever to perform experiments at
convincing scales. However, the number of whole-
ecosystemexperiments has declined in recent years.
Colleagues in the USA and Europe, as well as
elsewhere in Canada, report that proposals for
whole-ecosystem experiments have often been re-
jected by granting agencies because of concerns
about replication or high costs. The problem has
been exacerbated by investigatorswho havemisrep-
resented replicated mesocosms as being ‘‘ecosystem
scale.’’ This may effectively be the case in certain
areas that have been highlymodified by humans, by
eliminating or restricting the movements of large,
free-ranging predators that once had important
influences on community structure and ecosystem
function. Indeed, the highly simplified ecosystems
of plot-scale experiments have been the backbone
of modern agricultural development (Pimentel and
others 1992). In my experience, however, plots or
enclosures usually poorly simulate ecosystems with
more or less natural assemblages of species at
several trophic levels, and withmost of their natural
functions intact. Whether they are lakes with sev-
eral species of fish, rivers with migrating fish, or
terrestrial systems with large, free-ranging carni-
vores, important aspects of ecosystem and commu-
nity functions and responses to perturbation are
controlled by ‘‘keystone’’ organisms too large or
mobile to confine in experiments that are smaller
than ecosystem scale. Below, I shall make the case
that replicated or not, ecosystem-scale experiments