Direct Impacts of Climate Change on Freshwater Ecosystems - Climate Change Impacts on Freshwater Ecosystems - page 52

date of freezing and break-up, both advanced by about 6 days per 100 years
(Magnuson et al . 2000). This prolongation of the ice-free season translates to a
1.2°C increase in air temperature per 100 years. Inter-annual variability in the ice
phenology data from the selected rivers and lakes has increased since the 1950s. At
Lake Baikal, the ice-free season has lengthened by about 16 days per 100 years,
with a more consistent trend towards later freezing despite variability from decade
to decade. The trend towards earlier break-up occurred mainly before 1920
(Livingstone 1999; Magnuson et al . 2000; Todd & Mackay 2003) and reflects the
close relationship of melting to the April mean air temperature, which has shown
no trend since the 1920s, in contrast to other months of the winter. In Canada,
most lakes have experienced earlier break-up and a longer ice-free season during
the past decades (e.g. Futter 2003; Duguay et al . 2006). The trends in lake ice-
cover provide further evidence of a spring warming that has been observed over
North America since the second half of the 20th century. Ice phenology data from
Lake Mendota (Wisconsin, the United States), the lake with the longest uninterrupted
ice record in North America, dating back to 1855, showed three climatic periods
with relatively stable ice conditions between 1890 and 1979 but decreasing trends
in ice-cover duration before and after this period due to increasing winter/early
spring air temperatures (Robertson et al . 1992).
Long series of lake ice data are available from Fennoscandia. Korhonen (2006)
analysed freezing and break-up records of almost 90 lakes in Finland dating back
to the early 19th century and ice thickness records of about 30 lakes dating back
to the 1910s. There were significant changes with earlier ice break-up, except for
the very north of Finland, and later freezing, resulting in shorter ice duration.
Palecki & Barry (1986) conducted a statistical study of the relationship between
ice phenology data and air temperature for Finnish lakes, and found that the
same change in freeze-up date indicated a larger shift in autumn air temperature
in the north of Finland than in Southern Finland. Ice phenology dates revealed a
strong dependence on latitude, and the maritime influence of the Baltic Sea
caused a northward deflection of average freezing and melting date isolines near
the coast. A similar dependence of ice phenology on latitude was described by
Blenckner et al . (2004) for 50 lakes in Fennoscandia. The prevalence of zonal or
meridional winds in autumn and spring, expressed by regional circulation indices,
was used to explain the temporal and regional variability of freeze-up and
break-up dates.
An analysis of 54 Swedish lakes (Weyhenmeyer et al . 2005) confirmed the
existence of trends in the timing of melting during the IPCC reference period
(1961-90), but also showed that these trends were dependent on latitude. Trends
towards earlier melting were substantially greater in warmer, southern Sweden
than in the colder, northern regions. The non-linear dependence of break-up
dates on latitude can be described in terms of an arc cosine function of lake-
specific annual mean air temperature for 196 lakes across Sweden (1961-2002)
(Weyhenmeyer et al . 2004). Thus, an increase in air temperature is expected to
have greater impacts on the melting date in warmer southern Sweden compared
with the colder north with mean annual air temperatures of −2°C to 2°C. The
authors showed that the average temperature increase of 0.8°C from 1991 to
2002 across Sweden (compared with the reference period 1961-90) caused ice