eLearning

deliberately designed to test the applicability of ELA
results to larger systems show that, for some lake
processes, scaling can be very important.
The Northern Ontario Lake Size Study (NOLSS)
(Fee and Hecky 1992) examined changes in commu-
nities and in-lake processes in northwestern On-
tario lakes over a size range of several orders of
magnitude, ranging from ELA lakes of 4–55 ha to
Lakes Nipigon and Superior. Up to 10 years of data
were collected on some of the lakes. As lakes
increase in size, advective mixing processes, most
notably internal waves in the region of the thermo-
cline, become increasingly important in determin-
ing the exchange of nutrients and other chemicals
between epilimnion and hypolimnion. However,
less obvious factors were also functions of size. Fish
in smaller lakes tended to have higher mercury
concentrations than in larger lakes, because the
ratio of methylation to demethylation is a function
of lake depth and temperature (Bodaly and others
1993; Ramlal and others 1993). Nutrient status and
nutrient availability also depended on lake size (Fee
and others 1994; Guildford and others 1994). Ther-
mocline depth in ELA-size lakes is strongly affected
by the attenuation of light by dissolved organic
carbon (DOC) (Schindler and others 1996b), but in
larger lakes DOC becomes less important as physical
dynamics override the effect of light penetration on
thermocline depth (Fee and others 1996).
SUMMARY
In summary, many microcosm and mesocosm ex-
periments at ELA have yielded results that would
have caused erroneous management decisions, be-
cause responses were different from those of whole
lakes. In many cases, the problem was caused by
inadequate or erroneous scaling of sediment–water
interactions, physical phenomena, water renewal
times, and temporal events, which can often be
corrected. However, many of the faults are uncor-
rectable. In particular, the inclusion of rare but
important top predators and highly motile species is
impossible. I conclude that accurate ecosystemman-
agement decisions cannot be made with confidence
unless ecosystem scales are studied. Even at ecosys-
tem levels, attention to proper scaling enables more
accurate conclusions to be made, as the NOLSS
project has shown and Pace (forthcoming) has
concluded by comparing other studies. Design of
smaller-scale experiments to account properly for
the different scales of physical, chemical, and biologi-
cal processes is at least as important as replication. In
many cases, proper scaling will be impossible, and
experiments at less than ecosystem scales are inap-
propriate for making predictions about whole-lake
responses