eLearning

A wide variety of management models, including
those for nutrients, alkalinity generation, contami-
nant flushing, and conservative elements, are based
on assumptions of first-order kinetics between well-
mixed lake waters and sediments [for example, see
Vollenweider (1969, 1976), Dillon and Rigler (1974),
Ahlgren (1977), Schindler and others (1978), Baker
and others (1986), and Kelly and others (1987)].
When either water-flow rates or chemical influx
rates to such a systemare changed, chemical concen-
trations approach new steady-state asymptotes ex-
ponentially. As a rule of thumb, the system requires
at least three water renewal times to approach a
new steady state (Riggs 1963).
For most management purposes, the long-term
condition is of primary interest. Most lakes have
water renewal times ranging from several months
to over a decade. It would be difficult to use
mesocosms to predict steady-state chemical condi-
tions, unless complex models are used to design
chemical input regimes to simulate steady-state
conditions in lakes [for example, see Holoka and
Hunt (1996)].
Even if chemical steady state can be well simu-
lated, many organisms cannot respond to a treat-
ment in the few months that are usually the limit
for reliable mesocosm experiments. In northern
waters, most invertebrates have life cycles requiring
1–3 years, so even experiments encompassing most
of the ice-free season have low predictive power. As
an example, following introduction of predatory
Hesperodiaptomus arcticus to an alpine lake, the
zooplankton community has continued to change
for 5 years. There is still no evidence that Hesperodi-
aptomus has reached steady state (A. S. McNaught
and others unpublished; D. W. Schindler unpub-
lished).
There are even greater problems with larger
organisms. As Carpenter and Kitchell (1988) noted,
predator populations respond slowly tomost pertur-
bations, making it impossible to gauge accurately
their effects on other components of communities
in experiments that last only a few months, regard-
less of spatial scales. For example, fishes at the top of
the food chains in ELA lakes required at least 8 years
to respond fully to nutrient addition and acidifica-
tion, as well as to the removal of these stresses
(Mills and Chalanchuk 1987;Mills and others 1987;
Schindler and others 1993). Whole-ecosystem ex-
periments of short duration would have the same
deficiency.
Biodiversity can also be a problem in mesocosms,
for they may contain only a part of the total species
assemblage in a lake. Many whole-lake responses
occur either because of the surprise appearance of
organisms that are too rare to detect prior to
treatment or invade the lake fromelsewhere follow-
ing treatment. For example, after Lake 223 had
been acidified to pH 5, chironomid emergence was
undiminished. However, there was a dramatic
change in species. Over 95% of emergence was by
three species of the genus Cladotanytarsus, none of
which had been recorded in the lake in several years
of sampling at higher pH values. One of the species,
Cladotanytarsus aeiparthanus, had never been re-
corded before (Schindler and others 1985; Bilyj and
Davies 1989)! Similarly, following elimination of
dominant crustaceans fromSnowflake Lake by intro-
duced nonnative salmonids, a succession of crusta-
cean and rotifer species either rare or unrecorded in
the lake occurred over the next several years
(A. S. McNaught and others unpublished). In short,
isolating small parts of the ecosystem from the
whole may limit the range of responses possible in
mesocosms.
On the other hand, mesocosms may attract un-
wanted ‘‘biodiversity.’’ Elsewhere, I have recounted
the mischief that otters, muskrats, and water birds
can wreak onmesocosms, which offer nice perching
sites and containerswhere prey cannot easily escape
(Schindler 1988). Crayfish burrow under the walls
of mesocosms, entering and leaving them at will.
The above is not a complete list of problems with
mesocosms. Studies in other areas have revealed
many others (Gachter 1979;Marshall andMellinger
1980; Carpenter 1996).
PRACTICAL CONSTRAINTS
TO REPLICATING WHOLE ECOSYSTEMS
We often considered replicating whole-ecosystem
nutrient experiments at ELA, for the costs of fertil-
izer and acid are not formidable, ranging from a few
hundred to ten thousand dollars per year for chemi-
cals to treat a single lake. However, even with 46
lakes set aside for experimental purposes, it is
difficult to find near-replicates. Lakes differ in fauna
(Hamilton 1971; Patalas 1971; Beamish and others
1976); water renewal times, which determine the
rate of response to chemical changes (Schindler and
others 1978); and chemical concentrations (Arm-
strong and Schindler 1971). In the case of eutrophi-
cation experiments, we chose to use a covariance
approach, where phosphorus loadings and water
renewal times were different for every lake treated,
to say nothing of interannual differences in water
renewal (Schindler and others 1978). We treated
each year of each treatment as a separate experi-
ment, which is, of course, not true replication.
However, phosphorus renewal times are only a few