Aquat Ecol (2009) 43:653–659 DOI 10.1007/s10452-009-9284-1
Do warmer winters change variability patterns of physical and chemical lake conditions in Sweden? Gesa A. Weyhenmeyer
Received: 21 May 2009 / Accepted: 4 August 2009 / Published online: 18 August 2009 Ó Springer Science+Business Media B.V. 2009
Abstract In this study, the effect of a warmer winter climate on variability patterns of physical and chemical lake conditions was examined by using monthly air temperature data from 72 meteorological Swedish sites, ice breakup data from 77 Swedish lakes and monthly data of 17 water chemical variables from 11 nutrient-poor Swedish reference lakes during 1988–2005. The results showed significantly increasing variations of lake ice breakup dates and nitrate concentrations over Sweden along with increasing winter air temperatures. Variability patterns of other water chemical variables were not affected by warmer winters. Nitrate concentrations increased their variability in spring and early summer not only between lakes but also within lakes, which was attributed to a climate-induced increase in spring nitrate concentrations in particular in southern Sweden, while summer nitrate concentrations remained rather constant and low all over Sweden
G. A. Weyhenmeyer Department of Aquatic Sciences and Assessment, Swedish University of Agricultural Sciences, Box 7050, 750 07 Uppsala, Sweden G. A. Weyhenmeyer (&) Department of Ecology and Evolution/Limnology, Uppsala University, Norbyva¨gen 18D, 752 36 Uppsala, Sweden e-mail:
[email protected]
(median 10 lg l-1). Since nitrate concentrations play an important role for primary production, highly varying concentrations will be a challenge for biota to adapt. Keywords Ice cover Variability Nitrate Temperature North Atlantic Oscillation Lake
Introduction Climate change is one of the major concerns affecting Earth (IPCC 2007a, b). Considering a warming of the globe, drastic changes in geographical regions with snow and ice cover are expected (IPCC 2007a). In Sweden, snow and ice frequently occur during winter time, and the majority of the almost 100,000 Swedish lakes are ice covered during each winter (Weyhenmeyer et al. 2004). Since an ice cover has a strong impact on biogeochemical processes in lakes (e.g., Greenbank 1945; Weyhenmeyer et al. 1999; Ja¨rvinen et al. 2002), an increase in winter air temperatures is of particular importance for physical and chemical lake conditions in temperate and arctic regions. According to Kjellstro¨m (2004), a strong increase in winter air temperatures is expected for Sweden during the next decades. Winter air temperatures in Sweden are largely driven by the North Atlantic Oscillation (NAO), a climatic
123
654
phenomenon in the North Atlantic Ocean that represents fluctuations in the difference of atmospheric pressure at sea level between the Icelandic Low and the Azores High (Chen and Hellstro¨m 1999). Effects of an increase in the NAO and consequent warmer winter air temperatures on Swedish lake ecosystems have earlier been reported (Weyhenmeyer et al. 1999; Weyhenmeyer 2001, 2004; Blenckner et al. 2004), but these studies, like most other climate-related studies, focused on the assessment of mean values and trends of physical, chemical and biological lake variables. This kind of assessment, however, does not sufficiently address variability patterns of biogeochemical conditions despite the fact that they are increasingly seen as a critical factor driving the structure and function of aquatic ecosystems (e.g., Puckridge et al. 1998). Increases in the variability of biogeochemical conditions might foreshadow ecological regime shifts (Carpenter and Brock 2006; Guttal and Jayaprakash 2008) with possible catastrophic consequences (Scheffer and Carpenter 2003). Climate change is one possible process that has been suggested to result in variability changes of a variety of physical, chemical and biological variables (e.g., Walther et al. 2002; IPCC 2007a; Weyhenmeyer 2008). Variations of physical, chemical and biological lake variables occur on a spatial and temporal scale. A high variability on a spatial scale expresses that lakes are dissimilar in their biogeochemical conditions at a given time. Such dissimilarity between lakes can have many causes but, restricting the study to natural nutrient-poor reference lakes with very little human influence, possible changes in the dissimilarity in physical and chemical lake conditions over time are assumed to be strongly driven by climate. Focus in this study was on the impact of increasing winter air temperatures on variability patterns of physical and chemical lake conditions. Since winter air temperatures strongly affect the timing of ice breakup (Weyhenmeyer et al. 2004), it was hypothesized that the variability in the timing of ice breakup and ice breakup-related variables will increase between lakes along with increasing winter air temperatures. To test this hypothesis, variability patterns of ice breakup dates and water-chemistry data of 17 different variables from lakes distributed all over Sweden during the time period 1988–2005 were analyzed and related to time series of winter air temperature.
123
Aquat Ecol (2009) 43:653–659
Methods Complete data series of ice breakup dates from 77 lakes distributed over Sweden (Fig. 1) during 1988– 2005 were obtained from the Swedish Meteorological and Hydrological Institute (for further details on ice breakup data see Weyhenmeyer et al. 2004). For 11 lakes (Fig. 1), monthly water-chemistry data during the main growing season, i.e., May to October, were available from 1988 to 2005. These comprised data on pH, alkalinity (Alk), conductivity (Cond), calcium (Ca), magnesium (Mg), sodium (Na), potassium (K), chloride (Cl), sulfate (SO4), ammonium–nitrogen (NH4–N), nitrate–nitrogen (NO3–N), total nitrogen (TN), total phosphorus (TP), phosphate–phosphorus (PO4–P), absorbance at 420 nm of 0.45 lm filtered
Fig. 1 Map of Sweden showing locations of 77 lakes with available ice breakup dates (circles) and locations of 11 references lakes with available monthly water chemical data (dots)
Aquat Ecol (2009) 43:653–659
water in a 5-cm cuvette (Color), total organic carbon (TOC) and reactive silica (Si). All the 11 lakes have been classified by the Swedish National Protection Agency as nutrient-poor Swedish reference lakes with little or no anthropogenic changes other than climatic and atmospheric deposition and can be found in the national freshwater database at http://www.ma. slu.se (average nutrient concentrations of the reference lakes during 1988–2005: TP: 10 ± 5 lg l-1, TN: 402 ± 136 lg l-1, NO3–N: 54 ± 60 lg l-1) The catchment areas of the lakes are dominated by forest. The lakes are relatively shallow (mean lake depth ranging between 4 and 12 m) and small (lake surface area ranging between 0.16 and 1.70 km2). The monthly sampling was carried out in the middle of each lake at a depth of 0.5 m. NO3–N concentrations were determined by a photometric Bran & Luebbe autoanalyzer and included nitrite concentrations. This measurement approach is acceptable for our lakes since nitrite accounts for\5% of the sum of NO3–N and nitrite concentrations (Dr. A. Wilander, personal communication). All chemical analyses were performed by the same certified laboratory according to EN or ISO standards (Wilander 1998). Further information on methods and data is available at http://www.ma.slu.se. To evaluate the impact of winter air temperature changes on lake conditions, monthly air temperature data from 72 meteorological sites, representative of the lake sites distributed all over Sweden, were obtained from the Swedish Meteorological and Hydrological Institute for the time period 1988– 2005. Winter time was defined as the 3-month period from December to February. In addition to air temperatures, also the large-scale climatic index NAOw (North Atlantic Oscillation winter index) that has a strong impact on Swedish winter climatic conditions was used. This index has been described in detail by Hurrell (1995) and was downloaded from the homepage of the National Center for Atmospheric Research, US at http://www.cgd.ucar.edu/cas/ jhurrell/indices.html. All statistical analyses were performed in JMP, version 7.0.2. (SAS Institute 2007). If no further details are given, numbers are referred to as means ± standard deviations. In JMP, the Brown and Forsythe (1974) test was used to judge whether variances were equal or not. The test is appropriate for non-normal distributed data and shows the F-test
655
from an Anova in which the response is the absolute value of the difference of each observation (i.e., a physical or chemical variable) and the group median (i.e., a year). In the Brown–Forsythe test, the mean value of the absolute difference of each observation and the group median was used, and it is defined as the dissimilarity between lakes at a given time.
Results From 1988 to 2005, the mean winter (December to February) air temperatures of 72 sites distributed all over Sweden varied between -5.1 and ?1.1°C. The warmest winters during the 18-year study period were those in 1988/89 and 1989/90 (Fig. 2a). These corresponded to maximum values of NAOw (Fig. 2b) since 1864. In general, the NAOw was a suitable proxy for mean winter air temperatures (R2 = 0.49, P \ 0.001, n = 18), and it was also related to the mean lake ice breakup (R2 = 0.22, P \ 0.05, n = 18) over Sweden from 1988 to 2005. Because of the exceptionally warm winters in 1988/ 89 and 1989/90, the ice breakup in lakes over Sweden was exceptionally early (April 21 ± 42 days in 1989 and April 12 ± 38 days in 1990, Fig. 2c). In contrast, the lake ice breakup was very late (May 20 ± 16 days) in 1996, which was the coldest winter during the period from 1988 to 2005. The air temperature sine functions following the warmest and the coldest winters differed largely at the beginning of the years,
Fig. 2 Yearly mean values and standard deviations of winter air temperatures from 72 sites (a), the North Atlantic Oscillation winter index (NAOw, b) and the timing of ice breakup from 77 lakes (c) from 1988 to 2005
123
656
Aquat Ecol (2009) 43:653–659
Fig. 4 Relationship between winter air temperatures (mean values of 72 sites) and the dissimilarity of ice breakup dates between 77 lakes (crosses) and NO3–N concentrations between 11 lakes in May (squares) and in August (diamonds) from 1988 to 2005. Dissimilarity was defined by using the Brown– Forsythe test with year as the group median (see ‘‘Method’’). The year was then replaced by the measured winter air temperature of the year
Fig. 3 Seasonal variation (mean and standard deviation) in air temperature a from 72 sites and NO3–N concentration b from 11 lake sites over Sweden during the years 1989 and 1996. The grey rectangles in panel a show the variability of ice breakup dates in 1989 (dark grey) and 1996 (light grey). The variability pattern of ice breakup dates corresponds to the time period when air temperatures vary around 0°C
resulting in a large annual air temperature amplitude following the coldest winter in 1995/96, and a small one following the warmest winter in 1988/89 (Fig. 3a). In general, decreasing annual air temperature amplitudes in Sweden were well reflected by increasing winter air temperatures (R2 = 0.75, P \ 0.0001, n = 1296). Examining variability patterns of lake ice breakup over Sweden along with increasing winter air temperatures, i.e., using the Brown–Forsythe variable (see ‘‘Methods’’) for each year and replacing the year by the measured winter air temperature of the year, gave a clear increase in the dissimilarity of ice breakup dates between lakes along with increasing winter air temperatures (R2 = 0.60, P \ 0.001, n = 18; Fig. 4). In contrast, winter air temperatures did not become more dissimilar between sites along with increasing winter air temperatures (P [ 0.05). Relating ice breakup dates to spring (mid-May when the 11 study lakes were ice-free), concentrations of the 17 water chemical variables for each
123
individual lake (altogether 187 correlations) revealed significant relationships in the majority of the lakes for Na, Cl, pH and NO3–N during 1988–2005 (nonparametric Spearman-rho test, P \ 0.05, n = 18 for each correlation, Table 1). While Na, Cl and pH generally decreased in May with an earlier ice breakup, NO3–N was the only variable that significantly increased in its concentration in most of the lakes in May during the years when the ice breakup was early (Table 1). Despite increasing NO3–N concentrations in May, summer (August) NO3–N concentrations remained rather constant and low (median: 10 lg l-1; mean: 50 lg l-1), resulting in a large intra-annual range of NO3–N concentrations when winters had been warm (Fig. 3b). Since NO3–N concentrations in the Swedish lakes decreased largely in response to decreasing NO3–N atmospheric deposition (Weyhenmeyer et al. 2007), all NO3–N concentrations were also de-trended over time by a linear function. The de-trended NO3–N data were even better related to the timing of lake ice breakup in all lakes using the Spearman-rho test described above. Increased winter air temperatures did not only result in an increase in the intra-annual variability of NO3–N concentrations within lakes, but also in an increase in the variability of NO3–N concentrations between lakes here referred to as the dissimilarity between lakes (Fig. 4). The dissimilarity increase in NO3–N concentrations between lakes along with increasing winter air temperatures corresponded to the dissimilarity increase in ice breakup dates
Aquat Ecol (2009) 43:653–659
657
Table 1 Nonparametric Spearman-rho test for relationships between the timing of lake ice breakup and 17 different water chemical variables, measured in May, for 11 Swedish lakes during 1988–2005 Variable
Median r
Minimum r
Maximum r
NO3–N
-0.48*
-0.67*
-0.01
Si
-0.26
-0.44
-0.13
SO4
-0.13
-0.34
0.28
NH4–N
-0.11
-0.35
0.11
TN
-0.03
-0.31
0.37
Color TOC
-0.01 -0.01
-0.14 -0.12
0.33 0.20
Ca
-0.01
-0.10
0.28
K
0.05
-0.21
0.40
TP
0.10
-0.07
0.44
Mg
0.13
-0.20
0.33
Cond
0.18
-0.10
0.34
PO4–P
0.19
0.00
0.43
Alk
0.19
-0.40
0.53*
pH
0.45*
0.20
0.46*
Cl
0.45*
0.17
0.64*
Na
0.58*
-0.05
0.73*
Shown are median, minimum and maximum r-values of the 11 lakes. Asterisks indicate significant relations (P \ 0.05). For abbreviations see ‘‘Methods’’
between lakes (Fig. 4). An increase in the dissimilarity in NO3–N concentrations between lakes with increasing winter air temperatures was still detectable in August (P \ 0.05; Fig. 4) but not thereafter (P [ 0.05). Both changes in ice breakup dates and NO3–N concentrations were more pronounced for lakes located in the southern part of Sweden. No other water chemical variable showed an increase in the dissimilarity between lakes in May or later in the year with increasing winter air temperatures (P [ 0.05).
Discussion Winter air temperatures in Sweden are known to be influenced by the large-scale North Atlantic Oscillation (e.g., Chen and Hellstro¨m 1999) and accordingly also ice breakup dates show a relation to the NAOw (Weyhenmeyer et al. 1999; Blenckner et al. 2004). This is confirmed by this study during 1988–2005 on the Swedish lakes, where relationship between the NAOw and the timing of ice breakup was significant.
The NAOw reached exceptionally high values in 1988/89 and 1989/90 (Fig. 2b), the years when ice breakup dates over Sweden were both very early and variable (Fig. 2c). The high dissimilarity between lakes in 1988/89 and 1989/90 could not be explained by dissimilar winter air temperatures since there is no increase in the dissimilarity of winter air temperatures between sites when winters are warm (P [ 0.05). It is suggested that the observed increase in the dissimilarity of ice breakup dates between lakes along with increasing winter air temperatures (Fig. 4) is caused by changing intra-annual temperature amplitudes in a warmer climate. Shortening of the winter season by a warmer winter climate results in a smaller air temperature amplitude, and a consequent increase in the variability of air temperatures around 0°C (Fig. 3a) when ice usually disappears from lake waters (Weyhenmeyer et al. 2004). Such a temperature amplitude effect on ice breakup variability patterns gives an explanation for the observed increased dissimilarity in ice breakup dates over Sweden when winter air temperatures increase (Fig. 4). The timing of ice breakup clearly depends on the air temperature sine function (Weyhenmeyer et al. 2004). The chemical lake variable that best follows this sine function, with clearly pronounced seasonal variations, is NO3–N (Pettersson et al. 2003; Khalili and Weyhenmeyer 2009; Fig. 3b). The shape of the sine function strongly depends on winter and spring NO3–N concentrations, as NO3–N concentrations during summer are usually much lower and often depleted (Weyhenmeyer et al. 2007; Fig. 3b). For shallow lakes, comparable with the lakes of this study, it has earlier been described that the relative NO3–N decrease from spring to summer significantly increases with increasing spring NO3–N concentrations as denitrification seems to be favored (Weyhenmeyer et al. 2007). A large annual amplitude of NO3– N concentrations implies a high intra-annual variability within lakes (Fig. 3b). No other water chemical variable showed such a high intra-annual variability. Changes in the intra-annual variability of NO3–N concentrations were most pronounced in the southern part of Sweden, probably because in that region also most pronounced changes in ice breakup dates had occurred. Because of this unequal response over Sweden, an increasing dissimilarity in NO3–N concentrations over Sweden along with warmer winter air
123
658
temperatures was observed (Fig. 4). First in August, NO3–N concentrations became low and rather similar over the entire country. Consequently, the effect of warmer winters on the dissimilarity of NO3–N concentrations between lakes was no longer detectable in August or thereafter (Fig. 4). NO3–N was the only water chemical variable for which the variability increased with increasing winter air temperatures, probably because it was the only variable in the nutrient-poor Swedish reference lakes that showed significantly increasing concentrations together with an earlier ice breakup. Increased winter/ early spring NO3–N concentrations in relation to warmer winters have earlier been explained by increased loading from the catchment (e.g., Weyhenmeyer 2004). Other variables responded more lakespecific to an earlier ice breakup or their values decreased at early ice breakup (Na, Cl, pH), resulting in a decreased variability between lakes in spring. Also phosphorus that might play an important role for NO3–N variations due to the regulation of photosynthesis and respiration in lake ecosystems (Weyhenmeyer et al. 2007) remained rather low and constant both within and between years in the 11 reference lakes during 1988–2005. In this study, it was shown how far a warmer winter climate can change variability patterns of lake ice breakup dates and ice breakup-related variables, which seems to be mainly NO3–N for the nutrientpoor Swedish reference lakes. The increase in the dissimilarity of ice breakup dates and NO3–N concentrations between lakes in a warmer climate was attributed to changes in seasonal variability patterns. Analyses of such variability patterns have so far received much less attention in the context of climate change than more aggregated measures of change such as changes in mean values or trends. This is unfortunate because for biota often seasonal maxima and minima are more important than their averages (e.g., Parmesan et al. 2000). A high variability of NO3–N concentrations is especially relevant as too high concentrations can cause water pollution, eutrophication or surface-water acidification, while too low concentrations might stimulate nitrogen-fixing cyanobacterial blooms in fresh and brackish waters (e.g., Vitousek et al. 1997; Rabalais 2002). Even though the nutrient-poor Swedish reference lakes are not subjected to these kinds of problems, they are good study objects to demonstrate the impact of
123
Aquat Ecol (2009) 43:653–659
global warming on variability patterns of physical and chemical lake conditions, which then can be used to relate to other factors driving variability patterns in more eutrophied lakes. During warm winters, air temperatures did not become more dissimilar between sites. Despite this, in the warmer climate, ice breakup dates and NO3–N concentrations in lakes became increasingly dissimilar. Taking into account that simulations show higher variability also for air temperatures in future (IPCC 2007a), an even more pronounced increase in the dissimilarity between lake physical and chemical conditions is expected in future. Such an increase in dissimilarity will represent a serious challenge to adaptation and mitigation strategies for lake ecosystems outlined in the IPCC (2007b) report. Acknowledgments The author of this work is a research fellow of the Royal Swedish Academy of Sciences supported by a grant from the Knut and Alice Wallenberg Foundation. Funding for this study was also received by the Swedish Research Council. Many thanks go to the editor and two anonymous reviewers for fruitful comments and to the Swedish Environmental Protection Agency as well as the laboratory of the Department of Aquatic Sciences and Assessment for financing, sampling and analyzing the water samples.
References Blenckner T, Ja¨rvinen M, Weyhenmeyer GA (2004) Atmospheric circulation and its impacts on aquatic ecosystems in Scandinavia. Boreal Environ Res 9:371–380 Brown MB, Forsythe AB (1974) Robust tests for the equality of variances. J Am Stat Assoc 69:364–367 Carpenter SR, Brock WA (2006) Rising variance: a leading indicator of ecological transition. Ecol Lett 9:308–315 Chen DL, Hellstro¨m C (1999) The influence of the North Atlantic Oscillation on the regional temperature variability in Sweden: spatial and temporal variations. Tellus Series A-Dyn Meteorol Oceanogr 51:505–516 Greenbank J (1945) Limnological conditions in ice-covered lakes, especially related to winterkill of fish. Ecol Monogr 15:343–392 Guttal V, Jayaprakash C (2008) Changing skewness: an early warning signal of regime shifts in ecosystems. Ecol Lett 11:450–460 Hurrell JW (1995) Decadal trends in the North Atlantic Oscillation: regional temperatures and precipitation. Science 269:676–679 IPCC (2007a) Summary for Policymakers. In: Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller HL (eds) Climate change 2007: the physical science basis. contribution of working group 1 to the fourth assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge
Aquat Ecol (2009) 43:653–659 IPCC (2007b) Summary for policymakers. In: Parry ML, Canziani OF, Palutikof JP, van der Linden PJ, Hanson CE (eds) Climate change 2007: impacts, adaptation and vulnerability. contribution of working group ii to the fourth assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge Ja¨rvinen M, Rask M, Ruuhija¨rvi J, Arvola L (2002) Temporal coherence in water temperature and chemistry under the ice of boreal lakes (Finland). Water Res 36:3949–3956 Khalili MI, Weyhenmeyer GA (2009) Growing season variability of nitrate along a trophic gradient—contrasting patterns between lakes and streams. Aquat Sci 71:25–33 Kjellstro¨m E (2004) Recent and future signatures of climate change in Europe. Ambio 33:193–198 Parmesan C, Root TL, Willig MR (2000) Impacts of extreme weather and climate on terrestrial biota. Bull Am Meteorol Soc 81:443–450 Pettersson K, Grust K, Weyhenmeyer G, Blenckner T (2003) Seasonality of chlorophyll and nutrients in Lake Erken— effects of weather conditions. Hydrobiologia 506–509: 75–81 Puckridge JT, Sheldon F, Walker KF, Boulton AJ (1998) Flow variability and the ecology of large rivers. Mar Freshw Res 49:55–72 Rabalais NN (2002) Nitrogen in aquatic ecosystems. Ambio 31:102–112 SAS Institute (2007) JMP statistics and graphics guide, version 7.0.1. SAS, Cary Scheffer M, Carpenter SR (2003) Catastrophic regime shifts in ecosystems: linking theory to observation. Trends Ecol Evol 18:648–656
659 Vitousek PM, Aber JD, Howarth RW, Likens GE, Matson PA, Schindler DW, Schlesinger WH, Tilman DG (1997) Human alteration of the global nitrogen cycle: sources and consequences. Ecol Appl 7:737–750 Walther G-R, Post E, Convey P, Menzel A, Parmesan C, Beebee TJC, Fromentin J-M, Guldberg OH, Bairlein F (2002) Ecological responses to recent climate change. Nature 416:389–395 Weyhenmeyer GA (2001) Warmer winters: are planktonic algal populations in Sweden’s largest lakes affected? Ambio 30:565–571 Weyhenmeyer GA (2004) Synchrony in relationships between the North Atlantic Oscillation and water chemistry among Sweden’s largest lakes. Limnol Oceanogr 49:1191–1201 Weyhenmeyer GA (2008) Water chemical changes along a latitudinal gradient in relation to climate and atmospheric deposition. Clim Change 88:199–208 Weyhenmeyer GA, Blenckner T, Pettersson K (1999) Changes of the plankton spring outburst related to the North Atlantic Oscillation. Limnol Oceanogr 44:1788–1792 Weyhenmeyer GA, Meili M, Livingstone DM (2004) Nonlinear temperature response of lake ice breakup. Geophys Res Lett 31:L07203. doi:10.1029/2004GL019530 Weyhenmeyer GA, Jeppesen E, Adrian R, Arvola L, Jankowski T, Jennings E, Noges P, Noges T, Straile D (2007) Nitrate-depleted conditions on the increase in shallow northern European lakes. Limnol Oceanogr 52:1346–1353 Wilander A (1998) National lake survey 1995 (in Swedish with an English summary). Swedish Environmental Protection Agency, Stockholm, p 191
123