Skip booklet index and go to page content

Biodiversity in Canadian Lakes and Rivers

Trends in River and Lake Ice Break-Up/Freeze-Up

Return to Table of Contents

Page Contents

Ice cover plays a fundamental role in the biological, chemical, and physical processes that structure freshwater ecosystems (for example, Prowse, 2001a; Prowse, 2001b; Prowse and Culp, 2003; Huusko et al., 2007; Prowse et al., 2007b), in addition to their effects on artificial structures (for example, Jasek, 1998; Beltaos et al., 2006). The hydrological regimes of 58% of rivers in the Northern Hemisphere are affected by seasonal ice cover with major ice cover developing on 29% of these rivers (Prowse, 2005; Bennett and Prowse, 2010). Prowse (2005) reported that seasonal ice can develop as far south as 33°N in North America and 26°N in Eurasia, affecting 7 of the world’s 15 largest rivers and 11 of the 15 largest lakes (Prowse et al., 2007a). Canada is located north of 48°N and lake and river systems (in all ecozones+) are affected by ice ranging from periodic skim ice in the more southerly temperate regions to ice thickness in excess of 2 m in the high latitudes.

River ice is an integral component of the flow regime in cold region environments due to its hydraulic effect on water levels, its ability to restrict or alter flows in rivers, and to restrict gas exchange in lakes. Ice freeze-up, cover, and break-up can cause both direct (such as timing and magnitude of extreme hydrologic events such as low-flow and ice jam-flood events, for example, Beltaos et al., 2006) and indirect changes to the hydrological regime. These changes can significantly modify channel geomorphology (such as through bed scour), alter aquatic chemical processes, and affect aquatic ecological communities. In addition, ice cover disconnects both flowing and standing water in lakes and rivers from the atmosphere and limits the sun’s energy input, influencing key physio-chemical and biological processes (such as dissolved gas concentration and photosynthetic capacity). Despite the clear importance of ice processes to freshwater ecosystems, long-term biological monitoring data during the ice season are limited and few trend datasets are available.

Prowse and Culp (2003) provided a thorough review of the effects of ice on aquatic ecological communities. Aquatic communities are susceptible to any changes in the hydrological, cryospheric, and atmospheric regimes. In general, the life cycles of many aquatic organisms are directly and indirectly influenced by ice cover duration, water temperatures, and hydrological variability (see Table 6 for examples). Using data from 1991 to 1998 for an alpine region in western Norway, Borgstrøm (2001) found the annual growth rates of brown trout (Salmo trutta) were negatively correlated with spring snow depth. Results showed that during the deep snow years, 1992 to 1995, the observed average annual growth of the age-classes 6 to 8 reduced by about 50% compared to years with less snow in spring (1991 and 1996). Another study by Cunjak et al. (1998) demonstrated that inter-annual variability in the survival of juvenile Atlantic salmon in Catamaran Brook, New Brunswick, improved with increasing average winter flow but highest mortalities were associated with winter breakups and ice-jams triggered by rain-on-snow snowmelt events.

Top of Page

Table 6. Literature review summary of physical habitat changes and the direct and indirect effects on instream biodiversity and habitat availability in ice impacted rivers.

Table 6a. Ice freeze-up
Physical habitat changeImpacts and effects on instream biodiversityExample study
Decline in water temperatureLower metabolism (−)-
Decline in water temperatureDecline in food requirements (−)-
Decline in water temperatureReduced activity (−)-
Reduction in amount and quality of habitatRedistribution of juvenile fish to more desirable wintering habitat (−)> Rimmer et al. (1984)
Creation of new refugia (e.g., border ice)Protection from predation (+)-
Creation of new refugia (e.g., border ice)Low velocity refugia (+)-
Development of frazil iceOffers incubating medium for Atlantic tomcod (Microgadus tomcod) (+)Power et al. (1993)
Development of frazil iceAbrasion of gills (−)-
Development of frazil icePlugging gill rakers (−)-
Development of frazil iceMovement of fish away from preferred specific velocity habitat (−)-
Supercooled water/ development of anchor iceSignificant mortality of benthic invertebrates and fishes (eggs / juveniles) (−)Power > et al. (1993)
Supercooled water/ development of anchor iceIce in spawning habitats restricts oxygen to redds (−)-
Supercooled water/ development of anchor iceAlteration to flow regime leading to stranding / suffocation (−)Stickler et al. (2008)
Supercooled water/ development of anchor iceIncreased downstream drift on release of anchor ice (−)-
Supercooled water/ development of anchor iceAlteration in habitat use-
Decline in river flow/ water levelsosure of redds (−)Cunjak et al. (1998)
Table 6b. Main winter
Physical habitat changeImpacts and effects on instream biodiversityExample study
Increased ice in littoral zoneOrganisms migrate deeper (−)Li et al. (2007)
Increased ice in littoral zoneAdopt diapauses (−)-
Increased ice in littoral zoneAbility to overwinter in anchor ice (−)-
Increased ice in littoral zoneMortality from prolonged exposure to minimum temperatures (−)Finstad et al. (2004)
Increased ice in littoral zoneLower metabolic rate in juvenile Atlantic salmon-
Development of isolated pools causing habitat disruptionProvide overwintering habitat for some fish species (+)-
Development of isolated pools causing habitat disruptionUse of spring-fed habitat for overwintering of eggs and provision of food (+)-
Development of isolated pools causing habitat disruptionEffect on predator/prey relationships (−)-
Establishment of ice cavity habitatShoreline access for aquatic mammals (+)-
Establishment of ice cavity habitatWell insulated air cavities provide habitat (+)-
Establishment of ice cavity habitatStress for aquatic plants (−)-
Establishment of ice cavity habitatProlonged ice cover can reduce food sources, e.g., periphyton (−)-
Reduced dissolved oxygenIncreasing susceptibility to stress, predation and contaminants (−)-
Reduced dissolved oxygenSublethal effects, e.g., changes to cardiac and metabolic functions, reduced growth and swimming capacity (−)-
Reduced dissolved oxygenLocalized winterkills of fish from overcrowding in habitats (−)-
Table 6c. Ice break-up
Physical habitat changeImpacts and effects on instream biodiversityExample study
Scouring of river bedLow survival rates of eggs and juveniles (−)Cunjak et al. (1998)
Scouring of river bedLoss / shifts in of aquatic and riparian vegetation (−)Cameron and Lambert (1971)
Scouring of river bedMacroinvertebrate larvae can demonstrate avoidance behaviour by using the substrate as refugia (−)Scrimgeour et al. (1994)
Transport of large bedload materialLoss of habitat (−)Cunjak et al. (1998)
Transport of large bedload materialAffect survival rates of eggs and juveniles (−)-
Transport of large bedload materialEffect highly dependent on timing of breakup (−)-
Ice jam floodingSustaining water levels in riparian ponds and wetlands (+)Prowse and Culp (2003)
Changing water levelsStranding and suffocation of fish (−)Needham and Jones (1959)
Cott et al. (2008)
Clague and Evans (1997)
High sediment levelsImproved productivity because of increased organic material (+)-
High sediment levelsReductions in species diversity and abundance through loss of quantity and quality of habitat (−)Elwood and Waters (1969)
High sediment levelsOften immediate effect on benthic invertebrates but lagged effect in fish (−)-

Top of Page

Previous analyses of variability in ice cover on Canadian freshwater lakes and rivers have been limited in geographic scope, focusing on phenomenological metrics, such as timing of autumn freeze-up or spring break-up (for example, Williams, 1970; Brimley and Freeman, 1997; Jasek, 1998; Lacroix et al., 2005). The date of freeze-up is defined as the first day on which a water body is observed to be totally ice covered, while the break-up date is the day of the last ice observed, before the open-water phase. To date, trends research in river and lake ice break-up and freeze-up has been limited by a lack of long-term observations of consistent, objectively defined parameters. Within Canada, monitoring of ice cover began in 1822 in Toronto Harbour, Lake Ontario, with additional sites added to the network over time (Power et al., 1993; Wania and Mackay, 1993). Although the Canadian Ice Database contains 63,656 records covering the period from ice season 1822/1823 to 2000/2001, the network has dramatically declined in recent years, due to a lack of funding (Lenormand et al., 2002). For instance, the freeze-up/break-up network as of the 2000/2001 ice season represented only 4% of the 1985/1986 network (Lenormand et al., 2002). Recent developments in satellite imagery analysis (Brown and O'Neill, 2002) and the development of the national volunteer ice monitoring program, IceWatch (Environment Canada and Ice Watch, 2008), could allow the current network to be expanded and increase the spatial information for the largely inhabited areas of Canada.

A recent study explored long-term trends in lake ice data using records from as early as 1822 across Canada (Environment Canada and Ice Watch, 2008). The study combined data from IceWatch, a volunteer-run monitoring scheme, with data collected by the Meteorological Service of Canada and the Canadian Ice Service. Of the 950 sites within the database, nearly a third were represented by only one or two years of data (Environment Canada and Ice Watch, 2008). Limiting our analysis to sites with at least eight years of data, where the last observations were in or after 1990, a non-parametric Mann-Kendall time-series analysis demonstrated that 15 of the 195 sites showed significant trends towards earlier freeze-up (p<0.05) and an additional 15 sites demonstrated significant trends towards later freeze-up (p<0.05) (Environment Canada and Ice Watch, 2008). Trends were also reported in the timing of spring thaw, with 40 of the 258 sites demonstrating significant trends towards earlier melt (p<0.05) compared with only 5 showing significantly later melt (p<0.05) (Environment Canada and Ice Watch, 2008). Examination of the sites with non-significant trends shows that 168 of the 258 sites are demonstrating a tendency towards earlier spring melt compared with 75 sites showing a tendency towards later spring melt, although these trends were not statistically significant (Environment Canada and Ice Watch, 2008). Further examination of the results clearly shows that the rate of change of lake ice thaw was much more rapid from 1950 to the present as compared with the first half of the 20th century (Environment Canada and Ice Watch, 2008).

The record length for each site was variable within this analysis, however, and thus we revisited the data for this report to provide an additional analysis using consistent record lengths and time periods. We selected data for two separate analyses: i) 1970 to 2002, with a minimum of 25 years of data; and ii) 1900 to 2000, with a minimum of 80 years of data. These periods were chosen to maximize the number of sites for analysis while maintaining strict data quality controls. However, it should be noted that not all variables were available for all years, with a bias towards those related to ice break-up. In addition, although more recent data were available (up to 2007 within this dataset), there were no sites with longer-term data for more recent years. Our analysis followed the approach outlined in the IceWatch report (2008), in terms of variable selection and treatment. For the 1970 to 2002 period, only one out of 24 sites with suitable data demonstrated a significantly later date for freeze-up (p<0.05). The remaining sites did not demonstrate a tendency towards either earlier or later freeze-up. Examination of the break-up dates suggests eight out of 69 sites showed a statistically significant trend (p<0.05) for earlier break-up. Reflecting the results of the original IceWatch report, a large proportion of the remaining sites (46 out of 69 sites) showed a tendency towards earlier break-up with only 14 out of 69 sites demonstrating a tendency towards later break-up. Of the 14 available sites for analysis of the long-term data (1900 to 2000), only one site presented a statistically significant trend towards earlier break-up but this was at the 10% level. However, 10 of the 14 sites showed a (non-significant) tendency towards earlier break-up. The three sites with freeze-up data demonstrated statistically significant trends towards later freeze-up (p<0.01).

Despite lacking long-term data from a spatially extensive network, a study by Magnuson et al. (2000) reported consistent evidence of later freezing and earlier break-up (see example Canadian lakes and rivers in Table 7 and Table 8). The analysis on data from 39 lakes and rivers across the Northern Hemisphere from 1846 to 1995 demonstrated that freeze-up occurred on average 5.8 days later per century, and break-up occurred 6.5 days earlier per century. These results were likely largely the result of an increase in average air temperature of about 1.2°C per century. Three of the sites from Russia, Finland, and Japan had records extending back to the 18th century. These records suggest that trends were already apparent at that time, but that the trend rates continued to increase after 1850. For example, Lake Suwa, Japan, indicated freeze dates later by 2.0 days per century (p<0.0001) for the 550-year record. However, breaking the trend into time blocks demonstrated that freeze dates ranged from 3.2 days per 100 years (1443 to 1592) to 20.5 days per 100 years (1897 to 1993) (Magnuson et al., 2000). However, the study was geographically limited and thus it is inadvisable to draw wider conclusions at this time.

Although the continental-scale relationships indicate a significant decrease in ice cover in the Northern Hemisphere over the past 300 years, regional responses have demonstrated greater variability (see examples summarized in Table 7 for freeze-up and Table 8 for break-up). For example, a study by Williams (1970) found that the break-up on the Saint John River, New Brunswick, occurred 15 days earlier in the 1950s than in the 1870s. Also, the median break-up and freeze-up dates for the Red River at Winnipeg, Manitoba, were 12 and 10 days earlier and later, respectively, in the 19th century than in the 20th century (Rannie, 1983). Other regional and continental scale studies using observations that are more detailed have demonstrated strong patterns in freeze-up and break-up timing between decades and regions reflecting larger-scale atmospheric patterns (see examples summarized in Table 7 and Table 8). For example, Duguay >et al.. (2006) explored lake ice break-up and freeze-up trends for sites across Canada (example of lake break-up in Figure 25). Using the non-parametric Mann-Kendall trend analysis, they compared trends for three separate 30-year time periods: 1951 to 1980,  1961 to 1990, and 1971 to 2000 in addition to exploring 1966 to 1995. Their results demonstrated few clear spatial trends in lake freeze-up for the three periods (exceptions are presented in   Table 7) with the few significant spatial clusters suggesting more local or regional influences. These results are not consistent with those found by Zhang et al. (2001) who showed widespread trends towards earlier freeze-up across Canada but this may reflect the different spatial and temporal distribution of sites between these two studies.

Top of Page

Table 7. Summary of scientific studies quantifying trends in freeze-up for Canadian lakes and rivers using data up to and including the year 2000.
SiteLocationHabitatRecord years# yrs with data  
(# sites, if applicable)
Trend directionTrend significanceReference
Mackenzie RiverTaiga Shield, Boreal Shield, Boreal Plains, Montane Cordillera,
Taiga Plains, Taiga Cordillera, Boreal Cordillera
River1868–1978106.1 days later/100 years&lt;0.01Magnuson et al.(2000)
Red RiverSouthern Manitoba (unsure of ecozone+)River1799–19811613.2 days later/100 years&lt;0.001Magnuson et al.(2000)
Toronto HarbourMixedwood PlainsRiver1822–19201136.9 days later/100 years&lt;0.001Magnuson et al.(2000)
Red RiverBoreal Plains, PrairiesRiver1815–198115312 days later during 20th century Rannie (1983)
Frame LakeBoreal PlainsLake1956–1980250.4 days later/year&lt;0.1Duguay et al. (2006)
RHBN stations across CanadaCanadaRiver(a) 1967–1996
(b) 1957–1996
(c) 1947–1996
(a) 30Footnote a of Table 7(151)
(b) 40Footnote b of Table 7(71)
(c) 50Footnote b of Table 7 (47)
(a) earlier at 21.4% of sites(b) earlier at 38.2% of sites(c) earlier at 50.0% of sites&lt;0.1Zhang et al. (2001)
Grand LakeAtlantic MaritimeLake1952–1980290.58 days earlier/year&lt;0.1Duguay et al. (2006)
Lake AthabascaTaiga Shield, Boreal Shield, Boreal PlainsLake1965–1990231.25 days later/year&lt;0.01Duguay et al. (2006)
Deadman's PondNewfoundland BorealLake1961–1990280.5 days earlier/year&lt;0.05Duguay et al. (2006)
Lake UtopiaAtlantic MaritimeLake1971–2000301.23 days later/year&lt;0.001Duguay et al. (2006)
Island LakeBoreal ShieldLake1971–1998210.42 days earlier/year&lt;0.05Duguay et al. (2006)
Rivers across CanadaCanadaRiver(a) 1951–1980
(b) 1961–1990
(c) 1966–1995
(d) 1950–1998
(a) 30Footnote b of Table 7(50)
(b) 30Footnote a of Table 7 (68)
(c) 30Footnote a of Table 7 (60)
(d) 49Footnote a of Table 7 (41)
(a) 1.0 days later/decade
(b) 0.1 days earlier/decade
(c) 0.1 days later/decade
(d) 0.3 days later/decade
<0.1Lacroix et al.
(2005)

Top of Page

Notes of Table 7

Note a of Table 7

Each site does not necessarily have data for each year in the analysis

Return to note a referrer of table 1

Note b of Table 7

All sites have data for >2/3 of the years in the analysis

Return to note b referrer of table 1

Top of Page

Table 8. Summary of scientific studies quantifying trends in break-up for Canadian lakes and rivers using data up to and including the year 2002.
SiteEcozone+HabitatRecord years# yrs with data
(# sites)
Trend directionTrend significanceReference
Red RiverSouthern Manitoba (unsure of ecozone+)River1799–199318010.6 days earlier/100 years&lt;0.001Magnuson et al. (2000)
Toronto HarbourMixedwood PlainsRiver1822–19851117.4 days earlier/100 yearsNSMagnuson et al. (2000)
Miramichi RiverMixedwood PlainsRiver1822–19551277.3 days earlier/100 years&lt;0.01Magnuson et al. (2000)
Canoe LakeMixedwood PlainsLake1982–200117Mann-Kendall Z = -1.65&lt;0.05Futter (2003)
St. Nora LakeMixedwood PlainsLake1968–199021Mann-Kendall Z = -3.14&lt;0.001Futter (2003)
ScugogMixedwood PlainsLake1872–1995102Mann-Kendall Z = -1.73&lt;0.05Futter (2003)
SimcoeMixedwood PlainsLake1853–1995130Mann-Kendall Z = -1.82&lt;0.05Futter (2003)
StoneyMixedwood PlainsLake1956–198830Mann-Kendall Z = -2.30&lt;0.01Futter (2003)
Thirteen IslandsMixedwood PlainsLake1992–200110Mann-Kendall Z = -1.70<0.05Futter (2003)
Red RiverPrairie,
Boreal Plains
River1815–198115710 days earlier during 20th centuryRannie (1983)
Mackenzie River basinTaiga Shield, Boreal Shield, Boreal Plains and Cordillera, Montane Cordillera,
Taiga Plains and Cordillera
River1970–200233Footnotec(17)~1 day/decade in upstream basin< 0.1de Rham et al. (2008)
Yukon RiverBoreal Cordillera,
Taiga Cordillera
River1896–1998 -~5 days earlier/100 yearsJasek (1998)
RHBN stations across CanadaCanadaRiver(a) 1967–1996
(b) 1957–1996
(c) 1947–1996
a) 30Footnotec (151)
(b) 40Footnotec (71)
(c) 50Footnotec (47)
(a) earlier in 15.1% of sites
(b) earlier in 21.8% of sites
(c) earlier in 30.0% of sites
&lt;0.1Zhang et al. (2001)
Colpoys Bay (Lake Huron)Mixedwood PlainsLake1951–1980290.5 days later/year&lt;0.1Duguay et al. (2006)
Brochet Bay (Reindeer Lake)Boreal PlainsLake1951–1980300.5 days earlier/year&lt;0.05Duguay et al. (2006)
Gull LakeMixedwood PlainsLake1961–1990300.44 days later/year&gt;0.1Duguay et al. (2006)
Lake UtopiaAtlantic MaritimeLake1961–1990300.52 days earlier/year&lt;0.01Duguay et al. (2006)
Back Bay (Great Slave Lake)Taiga PlainsLake1971–1996260.4 days later/year&lt;0.05Duguay et al. (2006)
Diefenbaker LakePrairieLake1971–2000300.33 days earlier/year&lt;0.05Duguay et al. (2006)
Rivers across CanadaCanadaRiver(a) 1951–1980
(b) 1961–1990
(c) 1966–1995
(d) 1950–1998
(a) 30Footnoted (61)
(b) 30Footnoted (79)
(c) 30Footnotec (71)
(d) 49Footnotec (45)
(a) 1.0 days later/decade
(b) 2.2 days earlier/decade
(c) 2.0 days later/decade
(d) 1.6 days later/decade
&lt;0.1Lacroix et al. (2005)

Top of Page

Notes of Table 8

Note c of Table 8

Each site does not necessarily have data for each year in the analysis

Return to note c referrer of table 1

Note d of Table 8

All sites have data for >2/3 of the years in the analysis

Return to note d referrer of table 1

Top of Page

Conversely, trends in lake ice break-up reported by Duguay et al. (2006) suggest greater spatial coherence. Depending on the time period chosen for analysis, results suggest an earlier spring thaw in western Canada and a later spring thaw in eastern Canada (1951 to 1980). The 1961 to 1990 time period showed a nationwide trend towards earlier break-up and the  1971 to 1990 reflects that trend. Analysis of the 1966 to 1995 period reflects that of the 1961 to 1990 period with a general trend towards earlier break-up and few regional freeze-up trends. Previous studies (for example, Bonsal and Prowse, 2003; Bonsal et al., 2006) have demonstrated the link between ice break-up/freeze-up and the air temperature between one to three months before the event. As shown in Figure 25, the earlier trends in lake ice break-up follow the earlier arrival of the spring 0°C-isotherm date (Duguay et al., 2006). These results suggest a high degree of synchrony, with 78% of sites demonstrating a correlation (r&gt;0.5) between the isotherm date and the date of ice break-up.

Figure 25. Trends in lake ice break-up dates and spring temperature in Canada, 1966–1995. Source: UNEP (2007) with data from Duguay et al.(2006)
Figure 25 isa map showing trends in lake ice break-up dates and spring temperature in Canada, 1966–1995
Source: UNEP (2007) with data from Duguay et al. (2006)
Long description for Figure 25

This map of Canada depicts the areas of measured trends in lake ice break-up between 1966 and 1995 and shows the spring 0°C isotherm trend boundaries for 1966 to 1995. The map is marked with icons which indicate an earlier significant trend, an earlier non-significant trend, a later non-significant trend, or no trend. The map shows a general trend towards earlier break-up and few regional freeze-up trends.

Top of Page

Summary and future direction

Trends towards both later and earlier break-up and freeze-up across Canada have been reported in the scientific literature, but particularly towards earlier break up across Canada. Results from the IceWatch report (2008) suggested little evidence for trends in lake ice freeze-up but there were trends towards earlier lake ice break-up. These results were reflected in our analyses, particularly for the long-term data series. Within individual ecozones+, there was large variability. For example, data from stations in the Atlantic Maritime, Newfoundland Boreal, and Boreal Cordillera ecozones+ indicated earlier freeze-up trends in these ecozones+ while data from the majority of remaining regions indicated mixed trends. The majority of regions demonstrated earlier trends in ice break-up. There is a need to maintain and build on successful monitoring databases, such as IceWatch, in addition to developing GIS-based analyses of satellite imagery.

Top of Page

Return to Table of Contents