Abstract

In shallow lakes, environmental warming and nutrient loading are important influences on the likelihood of a shift between clear and turbid ecosystem states. With temperatures and nutrient runoff predicted to increase within the next decades, climate change poses a threat to lake communities. However, current predictions on the effect of these environmental factors on the abundance and timing of peak zooplankton numbers are based on correlations rather than on experimental isolation of thermal from other confounding effects. We present results of warming (4°C above ambient) and increased nutrient loading on plankton communities in 48 outdoor mesocosms, simulating fishless and hypertrophic ponds. The timing of the chlorophyll a peak and crustacean zooplankton peak abundance, dominated by Daphnia pulex, responded strongly to temperature and nutrient addition. Daphnia numbers reached peaks 22–24 days earlier in heated than in unheated mesocosms. The chlorophyll a peak abundance advanced by 15–19 days with heating. Phytoplankton, total zooplankton and D. pulex reached peak abundance 12–19 days later when doses of nitrogen and phosphorus were added; this finding contradicts predicted earlier phytoplankton and zooplankton spring peak abundances with nutrient enrichment. Peak zooplankton and D. pulex abundances did not differ with temperature treatment, contrary to our expectations, but peak abundances occurred at similar actual temperatures. Nutrient additions had no effect on the peak zooplankton and D. pulex abundances in our mesocosms. Overall, climate warming is likely to advance plankton phenology in fishless ponds; however, this advance could be dampened in systems with high nutrient concentration. We found very high zooplankton abundances with warming and high nutrient loadings inducing a clear water state in all our tanks owing to heavy zooplankton grazing despite high nutrient concentrations.

INTRODUCTION

Climate change is predicted to have huge impacts on the Earth's ecosystems through temperature increase, changed patterns of precipitation, more frequent extreme weather events and combinations of these (IPCC, 2007). In temperate regions, temperature is predicted to increase 3–5°C within the next century. Freshwater lake communities will be affected both by temperature and changing hydrology.

Shallow lakes and ponds are likely to be particularly susceptible to climate change (Mooij et al., 2005), as they have a high surface area compared with their volume and thus are very vulnerable to temperature increase. Previous research has shown that warming could exacerbate the impact of eutrophication and that the resulting population and ecosystem changes can include loss of fish due to oxygen starvation (McKee et al., 2003; Mooij et al., 2005, 2007; Feuchtmayr et al., 2009; Moran et al., 2010). Temperature increase has also been shown to favour the growth of toxic cyanobacteria in some eutrophic lakes, which may lead to problems for a range of organisms, including humans (Johnk et al., 2008).

Long-term data sets, sediment records, process-based modelling and space-for-time substitutions along latitudinal or other climatic gradients suggest earlier phytoplankton spring population increases with increasing temperature. Examples are: Müggelsee (Gerten and Adrian, 2000), Loch Leven (Carvalho and Kirika, 2003), three large Swedish Lakes (Weyhenmeyer, 2001), Lake Erken (Weyhenmeyer et al., 1999), Bassenthwaite Lake (Elliott et al., 2006), Plußsee (Müller-Navarra et al., 1997), Lake Washington (Winder and Schindler, 2004), Windermere (Thackeray et al., 2008) and Lake Constance (Peeters et al., 2007). For the deeper of these lakes, however, secondary hydrodynamic factors, stratification periods and mixing depths can play a major role in seasonal phytoplankton development (Diehl et al., 2002). Shallow lakes avoid these complications.

Further evidence of the influence of climate warming on freshwater systems can be found for zooplankton, a major component of freshwater food-webs, which influences water quality by consuming algae and providing food for fish. Zooplankters, mostly Daphnia spp., generally become very abundant in spring in temperate lakes following peaks in algal abundance. In North American lakes, the timing of maximum Daphnia abundances is linearly related to latitude, with earliest peaks in southern lakes (Gillooly and Dodson, 2000). Berger et al. (2007) found that while experimental cooling of lake enclosures did not affect the timing of the phytoplankton peak, it delayed the peak of Daphnia abundance. However, advances in the timing of the clear water period (presumed to follow the zooplankton peak in abundance), observed in Dutch lakes over 17 years with a water temperature increase of around 0.5°C (Scheffer et al., 2001), may be difficult to attribute wholly to effects of climate warming because of concurrent reductions in nutrient loadings (Van Donk et al., 2003).

An earlier peak abundance of zooplankton has been predicted in warmer years by Scheffer et al. (2001) in a single numerical application of a minimal model, with this model including predation by fish. More recently, Schalau et al. (2008) predicted, using a size-structured Daphnia population model, that temperature rather than food availability is likely to drive inter-annual variability of Daphnia population dynamics during spring: they specifically predicted an advanced timing of peak abundance with warming. In their model, Daphnia grazing and population growth were temperature dependent, but phytoplankton growth was not. In the present experimental study in systems that lacked fish, we predict that peak abundance should occur earlier in response to warming because the growth and grazing of zooplankton is generally more thermally sensitive than in primary production (Allen et al., 2005; López-Urrutia et al., 2006; Rose and Caron, 2007; Bissinger et al., 2008). Another idea, which derives from equilibrium models of populations at carrying capacity (Savage et al., 2004), but which has not yet been explored in models of transient dynamics, is that an increased per capita grazing rate relative to rate of algal production at increased temperatures may reduce the peak standing crop per mass of algal food.

Timing of peak plankton abundances is also affected by many other influences on grazer–prey dynamics, including their respective initial abundances at the end of winter, light and nutrients (e.g. Jäger et al., 2008; Thackeray et al., 2008). Trends towards heavier precipitation events with increased frequency will result in altered nutrient runoff into freshwater systems (Jeppesen et al., 2009). Thackeray et al. (2008) concluded that eutrophication and/or climate change induces phenological shifts in phytoplankton and that effects are species dependent. The experimental approach used in the present study can avoid or account for these confounding effects of temperature and nutrients. Experimental approaches are desirable but are expensive to set up and maintain and are therefore rare. Here we present results from a replicated experiment that manipulated temperature and nutrient availability in the absence of planktivorous fish. Fish-free systems are not only likely to become more common as a result of warming, they are already important components of many temperate landscapes as small ponds that are of particular conservation importance for invertebrates and amphibians (Biggs et al., 2005; Werner et al., 2007). Using fish-free systems also allowed elimination of a major top–down effect from the bottom–up effects.

In an earlier investigation using the same mesocosm tanks as used here, modest influences of a 3°C temperature increase were observed on phytoplankton, zooplankton, macroinvertebrates and macrophytes (McKee et al., 2002, 2003; Moss et al., 2003; Feuchtmayr et al., 2007). The former experiments used a temperature slightly above the upper limit of predictions made for a business-as-usual scenario by the Intergovernmental Panel on Climate Change in its Second Report published in 1995, on the grounds that there was a trend of increased warming in the predictions being made. These predictions were indeed revised upwards substantially in the Third Assessment of 2002 and the A2, business-as-usual, scenario predicted a 3°C mean global rise by 2100. For the current experiments, we thus chose a 4°C rise in line with the considered most likely A2 (regionally orientated economic development) predictions (2.0–5.4°C) over the 21st century. These predictions have again been raised in the Fourth Assessment Report of 2007. Experiments have hitherto not been performed on open air plant-free shallow systems at very high nutrient levels that simulate the sort of lake or pond that has already been strongly affected by eutrophication. With richer organic sediment than in the previous experiment (McKee et al., 2003), we ensured high levels of phosphorus released to the water column, and with realistic nitrogen loadings we could focus on plankton growth in turbid shallow lakes (Scheffer et al., 1993).

The current experiment explored the effects of 4°C warming and realistically high external nutrient loading on 8-fold replicated plant-free shallow systems during spring. It allowed us to separate temperature from nutrient impacts on zooplankton abundance and on any change in magnitude and timing of peak abundance. Our objectives were: (i) to test the influence of warming and nutrient loading on zooplankton abundance, (ii) to quantify the extent to which the timing of zooplankton peak abundance might be advanced by 4°C warming and nutrient additions, (iii) to test whether warming affects the peak abundance of zooplankton per unit chlorophyll, (iv) to test whether nutrient addition increases the peak abundance of zooplankton and (v) to describe the impacts of temperature and nutrients on shallow lake ecosystems.

METHOD

Experimental set-up and treatments

Forty-eight mesocosm tanks, each 1 m deep and 2 m in diameter, sunk in the ground (see McKee et al., 2000), are located at the University of Liverpool Horticultural and Environmental Research Station at Ness Botanic Gardens in North West England (53°16′N, 3°03′W), UK. In October 2005, the tanks were filled with a 20-cm deep sediment mixture containing (by volume) 50% garden loam and 50% organic material (47.5% chopped organic oat straw and 2.5% rotted organic cow manure to simulate a sediment of a eutrophic lake or pond), before being filled with around 3000 L of ground water pumped from a borehole.

A heating element was placed on top of the sediment either as a “dummy” in 24 non-heated tanks or connected to the heating system in 24 heated tanks. In heated tanks, 60°C hot water could be pumped through submerged pipes to increase the water temperature. Water temperature was measured continuously by sensors at a depth of 45 cm, and a computerized feedback system ensured a 4°C higher water temperature in heated tanks compared with adjacent non-heated tanks. Thus, heated tanks follow daily and seasonal temperature cycles of the ambient temperature tanks plus 4°C. Heating was switched on in October 2005 and tanks were allowed to settle until January 2006 when sampling and monitoring started.

Two different nutrient additions were superimposed on the heating treatments in a randomized block design from January 2006, via a standard nutrient addition every 2 weeks. Sixteen tanks did not receive nutrient additions (No), while the other two treatments, of 16 tanks each, received low (Low) and high (High) nitrogen loading (17.9 µM or 250 µg L−1 and 178.5 µM or 2500 µg L−1), each combined with a constant phosphorus loading (1.6 µM or 50 µg L−1) using sodium nitrate (NaNO3) and potassium dihydrogen orthophosphate (KH2PO4). Evaporation losses from the tanks were compensated with deionized water to avoid interference with nutrient levels and ionic composition. The study reported here represents the first, fishless phase (1st of January to 1st of June 2006) of a longer term experiment, which later included fish additions and macrophyte population establishment.

Inoculation

Zooplankton, phytoplankton and macroinvertebrates were collected from four ponds within the Botanic Garden and a nearby lake, Rostherne Mere, over several weeks and kept in a large storage tank. In January 2006, the storage tank contents were well mixed and sub-samples added to the 48 experimental mesocosms. In addition, D. magna was added to each mesocosm as ephippia from a nearby pond in October 2005, and 150 D. magna adults from the same origin on 20 March 2006. Two and 3 weeks after inoculation, invertebrates were cross-mixed among tanks via standardized sweep-net samples to ensure similar starting conditions.

Monitoring

Routine monitoring for water chemistry was based on Johnes and Heathwaite (1992) [total nitrogen (TN), total phosphorus (TP)] and Mackereth et al. (1989) [soluble reactive phosphorus (SRP), nitrate (NO3-N), ammonium (NH4-N)], in addition to conductivity, pH (by electronic meter), chlorophyll a (by a submersible fluorometer Cyclops-7, Turner Designs) and zooplankton counts took place every 2 weeks. Oxygen was measured at mid-water (35 cm) depth with an YSI Model 85 submersible system between 10.30 and 12.30 h BST (termed “mid-day”). Temperature was recorded automatically by the sensors and computer system every 15 min throughout the experiment. The phytoplankton community was examined microscopically to determine major components.

Zooplankton samples were taken after sunset to exploit a more equal distribution of the animals within the water column. Ten-litre integrated water samples were taken with a 1-m long tube, mixed in a bucket and concentrated for later transport to around 3 L with 100-µm mesh size gauze. Using the same method, the samples were concentrated down to 50 mL and fixed with sugar-formaldehyde the next morning. Sub-samples of at least 100 individuals (cladocerans and copepods) were counted under a microscope. Copepod numbers include copepodid stages. Nauplii were not included in the copepod counts nor were rotifers, nor Daphnia neonates containing egg yolk reserves.

To estimate the time difference caused by treatment (heated versus unheated and among nutrient additions) for the phytoplankton and zooplankton peaks, we applied two different methods. First, we recorded the date when the greatest number in each tank was reached; secondly, for an estimate between the fortnightly samplings, we fitted a curve (Weibull-type function) to data from each tank, providing a method for “peak” curve fitting following Rolinski et al. (2007), using the cardidates package of the statistical data analysis system R (R Development Core Team, 2007). Data shown for chlorophyll-a peak concentrations exclude seven mesocosms, distributed among the different treatments, where chlorophyll-a peaks developed before the start of our 2-weekly sampling regime preventing us from determining peak concentrations and fitting a curve to the data. For better comparison of the timing of zooplankton, D. pulex and phytoplankton peaks, results determined with fitted Weibull-type functions were standardized and Z-scores calculated (the difference in days between peak occurrence in each tank and the average of all tanks, divided by the overall standard deviation, and averaged for the different treatments). Hence, negative z-scores indicate earlier peaks compared with the average time of peak occurrence, positive values indicate later peaks.

Statistical methods

Wherever model residuals from analyses of untransformed data showed evidence of non-normality and heteroscedasticity, data were log-transformed before applying two-way ANOVAs with time as repeated measure (Table I). All analyses were examined for the influence of outliers by plotting Cook's distances. One-way ANOVAs were applied to test treatment effects on the timing of peak abundances (Table II) along with all other comparisons using SPSS 15.0 for Windows®. The same package was used for pairwise post hoc Bonferroni-corrected tests.

Table I:

Effects of heating (H) and nutrient (Nu) loading (no addition, No; addition of P and low addition of N, Low; addition of P and high addition of N, High) over time (t) on water chemistry, zooplankton abundance and chlorophyll a concentration, in 47 mesocosms from beginning of January 2006 to end of May 2006 (n = 10 for all parameters except where n is given)

 Mean value
 
Probability
 
 Unheated No Low High Nu H × Nu H × t Nu × t H × Nu × t 
Oxygen concentration (mg L−1) (n = 13) 4.8 2.4 3.2 2.9 4.8 *** *** ns *** ns ns 
Oxygen saturation (%) (n = 13) 40.5 21.9 27.6 24.8 41.6 *** *** ns *** ns ns 
SRP (µg L−181 64 65 78 73 ** *** *** ns ns 
TP (µg L−1319 326 257 349 357 ns *** ns ** ns ns 
Nitrate-N (µg L−183 45 10 12 168 *** *** ns *** ns 
Ammonium-N (µg L−1274 466 136 268 685 ns *** *** ** ns ns 
TN (mg L−12.1 2.3 1.6 1.9 3.1 *** *** ns *** ** ns 
Conductivity (µS cm−1782 884 826 830 840 *** ns *** ns ns 
Chlorophyll a (µg L−137 37 21 26 63 ns *** ns *** ns ns 
Zooplankton abundance (ind. L−1257 288 235 260 324 ** ns ns *** ns ns 
D. pulex abundance (Ind. L−1238 284 234 241 301 *** ns ns *** ns ns 
 Mean value
 
Probability
 
 Unheated No Low High Nu H × Nu H × t Nu × t H × Nu × t 
Oxygen concentration (mg L−1) (n = 13) 4.8 2.4 3.2 2.9 4.8 *** *** ns *** ns ns 
Oxygen saturation (%) (n = 13) 40.5 21.9 27.6 24.8 41.6 *** *** ns *** ns ns 
SRP (µg L−181 64 65 78 73 ** *** *** ns ns 
TP (µg L−1319 326 257 349 357 ns *** ns ** ns ns 
Nitrate-N (µg L−183 45 10 12 168 *** *** ns *** ns 
Ammonium-N (µg L−1274 466 136 268 685 ns *** *** ** ns ns 
TN (mg L−12.1 2.3 1.6 1.9 3.1 *** *** ns *** ** ns 
Conductivity (µS cm−1782 884 826 830 840 *** ns *** ns ns 
Chlorophyll a (µg L−137 37 21 26 63 ns *** ns *** ns ns 
Zooplankton abundance (ind. L−1257 288 235 260 324 ** ns ns *** ns ns 
D. pulex abundance (Ind. L−1238 284 234 241 301 *** ns ns *** ns ns 

Results are from a two-way ANOVA with time (t) as repeated measure.

Wherever results were significant, P-values are indicated as *P < 0.05, **P < 0.01, ***P < 0.001; ns indicates not significant results.

Table II:

Effects of heating (H) and nutrient (Nu) loading (no addition, No; low addition, Low; high addition, High) on timing of zooplankton peak abundance, Daphnia pulex peak abundance and chlorophyll a peak abundance, all calculated as Julian day with two different methods: (1) peak abundance sampled, and (2) fitting Weibull-type functions for 47 mesocosms from beginning of January 2006 to end of May 2006 wherever possible

 Mean value
 
Probability
 
 Unheated No Low High Nu H × Nu 
1) Time of zooplankton peak 138 114 121 127 130 *** ns ns 
2) Time of zooplankton peak 140 117 120 130 134 *** ns 
1) Time of D. pulex peak 138 114 119 128 131 *** ns 
2) Time of D. pulex peak 139 117 120 128 136 *** ns 
1) Time of chlorophyll a peak 54 39 45 41 53 ns ns 
2) Time of chlorophyll a peak 50 31 34 35 53 ns ns 
 Mean value
 
Probability
 
 Unheated No Low High Nu H × Nu 
1) Time of zooplankton peak 138 114 121 127 130 *** ns ns 
2) Time of zooplankton peak 140 117 120 130 134 *** ns 
1) Time of D. pulex peak 138 114 119 128 131 *** ns 
2) Time of D. pulex peak 139 117 120 128 136 *** ns 
1) Time of chlorophyll a peak 54 39 45 41 53 ns ns 
2) Time of chlorophyll a peak 50 31 34 35 53 ns ns 

ANOVA results are given, P-values are indicated as *P < 0.05, **P < 0.01, ***P < 0.001; ns indicates non-significant results.

RESULTS

Water temperature in ambient tanks increased from 4°C in January to about 18°C during the first half of May 2006, with the greatest change in April 2006, while the temperature in the heated tanks was 4 ± 0.1°C higher (Fig. 1). Throughout the period of observation, one heated mesocosm remained anaerobic with very low pH (by the end of May, pH mean ± SE 4.9 ± 0.17 compared with 7.1 ± 0.01 in the other 47 tanks); this outlier was not included in the analyses.

Fig. 1.

Mean daily water temperatures averaged over 24 ambient (open circles) and 23 heated (filled circles) mesocosms from January to end of May 2006.

Fig. 1.

Mean daily water temperatures averaged over 24 ambient (open circles) and 23 heated (filled circles) mesocosms from January to end of May 2006.

Mean oxygen concentration and saturation was significantly lower in heated tanks but increased with nutrient additions (Table I). Among nutrient treatments, oxygen concentration and saturation were significantly higher in the High treatment, but did not differ between No and Low treatments (post hoc test, P = 1.0). Phosphorus released from the sediment resulted in high TP concentrations (154–707 µg L−1). TP and SRP concentrations in both nutrient-addition treatments were higher than in tanks without experimental addition (Table I). The post hoc test showed a significant increase between No and Low treatments (P= 0.05) for TP, but no significant differences between other pairs.

TN concentrations were high, ranging from 1.35 to 4.42 mg L−1, significantly increasing with nutrient loading treatments (Table I). With warming, TN concentrations were high at the beginning of the experiment possibly due to release from the sediment, but decreased in February to the concentrations in unheated tanks, resulting in significant differences with temperature over time (Table I). Added nitrate did not persist long in the water and 2 weeks after addition was generally only detected in low concentrations except in the high nutrient treatment (Table I). Nitrate and ammonium concentrations increased with nutrient addition (Table I) and among nutrient treatments, pairwise comparisons between Low and High and No and High were significantly different. For TN, all post hoc results of nutrient treatments were significantly different.

The concentration of chlorophyll a was significantly higher in mesocosms with high nitrogen addition than in those with lower addition (pairwise Bonferroni-corrected comparison), but high variation among replicated tanks prevented detection of significant differences with heating (Table I, Fig. 2a). Phytoplankton biomass consisted mainly of Cryptophyceae (around 90%, averaged over all tanks), followed by Chrysophyceae (around 8%), and minor amounts of all other groups.

Fig. 2.

Mean ± SE of (a) chlorophyll a concentrations (µg L−1) and (b) crustacean zooplankton abundances (L−1) from end of January to end of May 2006 in heated and unheated mesocosms. Mean ± SE of (c) chlorophyll a concentrations (µg L−1) and (d) crustacean zooplankton abundances (L−1) from end of January to end of May 2006 in mesocosms with no nutrient addition, low nutrient addition (addition of P and low addition of N) and high nutrient addition (addition of P and high addition of N).

Fig. 2.

Mean ± SE of (a) chlorophyll a concentrations (µg L−1) and (b) crustacean zooplankton abundances (L−1) from end of January to end of May 2006 in heated and unheated mesocosms. Mean ± SE of (c) chlorophyll a concentrations (µg L−1) and (d) crustacean zooplankton abundances (L−1) from end of January to end of May 2006 in mesocosms with no nutrient addition, low nutrient addition (addition of P and low addition of N) and high nutrient addition (addition of P and high addition of N).

Crustacean zooplankton numbers consisted almost exclusively (84–96%) of Daphnia pulex. Daphnia magna appeared in May, but in low abundance (mean 1% of total zooplankton numbers). Cyclopoid copepods were relatively abundant until 4 April (mean 11% of total zooplankton numbers), but declined from 18 April onwards (mean 5%) as Daphnia pulex abundances increased. Non-crustacean zooplankton was dominated by protozoa with little representation of other groups. Total numbers of zooplankters and D. pulex alone increased with nutrient additions, increased significantly with warming and showed strong and significant responses to temperature over time when analysed by two-way ANOVA with time as repeated measure (Table I). Crustacean zooplankton abundance continuously increased in heated mesocosms from January to 18 April, and in unheated mesocosms from end of February to 16 May (Fig. 2b).

The timing of both measured and fitted peak abundances was significantly advanced with heating for total zooplankton, D. pulex-only and for chlorophyll a (see ANOVA results in Table II, Fig. 3). The measured mean peaks were 24 days, 24 days and 15 days earlier and the fitted mean peaks were 23 days, 22 days and 19 days earlier, respectively. There were also significantly delayed peaks with nutrient addition: 12 days (measured) and 16 days (fitted) for D. pulex and 14 days (fitted) for total zooplankton between No and High treatments (also see Fig. 2d). Non-significant delays in chlorophyll-a peak concentrations by the High treatment were 8 days (measured) and 19 days (fitted) (Table II, Fig. 2c). Z-scores in Fig. 3 show earlier peaks (negative values) for zooplankton, D. pulex-only and chlorophyll a with heating compared with later peaks (positive values) in ambient treatments, which were delayed with increasing nutrient addition.

Fig. 3.

Z-scores for chlorophyll a (black bars), total zooplankton (white bars) and D. pulex (diagonally lined bars) peak abundance determined with Weibull-type functions (based on Rolinski et al. (2007)), in 47 mesocosms from February to end of May 2006. Negative z-scores indicate earlier peaks compared with the average time of peak occurrence, positive values indicate later peaks (Treatments: UNo, Unheated and No nutrient addition; HNo, Heated and No nutrient addition; ULow, Unheated and Low nutrient addition; HLow, Heated and low nutrient addition; UHigh, Unheated and High nutrient addition; HHigh, Heated and high nutrient addition).

Fig. 3.

Z-scores for chlorophyll a (black bars), total zooplankton (white bars) and D. pulex (diagonally lined bars) peak abundance determined with Weibull-type functions (based on Rolinski et al. (2007)), in 47 mesocosms from February to end of May 2006. Negative z-scores indicate earlier peaks compared with the average time of peak occurrence, positive values indicate later peaks (Treatments: UNo, Unheated and No nutrient addition; HNo, Heated and No nutrient addition; ULow, Unheated and Low nutrient addition; HLow, Heated and low nutrient addition; UHigh, Unheated and High nutrient addition; HHigh, Heated and high nutrient addition).

To test whether warming or nutrient additions affect the peak abundance of zooplankton, we calculated mean plankton peak abundances for each treatment (Fig. 4). Peak abundances of zooplankton and D. pulex in heated tanks (mean ± SE 1170 ± 139 and 1158 ± 140) were higher than in unheated tanks (mean ± SE 940 ± 80 and 910 ± 84), but not significantly different (F1,45 = 2.02, P = 0.16, F1,45 = 2.28, P = 0.14 for total zooplankton and D. pulex, respectively, Fig. 4a). For nutrient treatments, zooplankton and D. pulex peak abundance was highest in tanks with high nitrogen addition (mean ± SE 1223 ± 174 and 1190 ± 170) compared with low addition (mean ± SE 952 ± 145 and 948 ± 158) and no addition (mean ± SE 992 ± 86 and 968 ± 86), but not significantly different between treatments (F2,44 = 1.08, P = 0.35, F2,44 = 0.87, P = 0.43 for total zooplankton and D. pulex, respectively, Fig. 4b). Chlorophyll a, however, measured on the morning after the zooplankton peak (sampled at night), was lower in the unheated mesocosms than in the heated mesocosms (ANOVA, F1,44 = 4.77, P = 0.03). Zooplankton per unit chlorophyll a at peak zooplankton abundance was similar in heated and unheated mesocosms (ANOVA, F1,43 = 1.71, P = 0.2 for zooplankton, F1,43 = 1.59, P = 0.22 for D. pulex, Fig. 4c) and in the different nutrient treatments (ANOVA, F2,42 = 2.58, P = 0.09 for zooplankton, F2,42 = 2.07, P = 0.14 for D. pulex, Fig. 4d) when measured on the morning after the zooplankton peak.

Fig. 4.

(a) Boxplots of total zooplankton and D. pulex peak abundances in heated and unheated mesocosms (F1,45 = 2.02, P= 0.16 for total zooplankton, F1,45 = 2.21, P= 0.15 for D. pulex) and (b) in mesocosms with no nutrient addition (No), low (Low) and high (High) addition (F2,45= 1.08, P= 0.35 for total zooplankton, F2,45 = 0.87, P= 0.43 for D. pulex). (c) Boxplots of peak zooplankton abundances (L−1) per chlorophyll a concentrations (µg L−1) 1 day after the zooplankton peak in heated and unheated mesocosms (F1,43= 2.97, P= 0.09 for zooplankton, F1,43= 1.80, P= 0.19 for D. pulex) and (d) in mesocosms with no nutrient addition, low (Low) and high (High) addition (F2,43= 2.58, P= 0.09 for zooplankton, F2,43 = 2.07, P= 0.14 for D. pulex). Two chlorophyll a readings were negative and were not included in the analysis; the large box represents data between the 25th and 75th percentile; the line within the box represents the median. Vertical bars show the upper and lower 10 percentiles, dots represent outliers.

Fig. 4.

(a) Boxplots of total zooplankton and D. pulex peak abundances in heated and unheated mesocosms (F1,45 = 2.02, P= 0.16 for total zooplankton, F1,45 = 2.21, P= 0.15 for D. pulex) and (b) in mesocosms with no nutrient addition (No), low (Low) and high (High) addition (F2,45= 1.08, P= 0.35 for total zooplankton, F2,45 = 0.87, P= 0.43 for D. pulex). (c) Boxplots of peak zooplankton abundances (L−1) per chlorophyll a concentrations (µg L−1) 1 day after the zooplankton peak in heated and unheated mesocosms (F1,43= 2.97, P= 0.09 for zooplankton, F1,43= 1.80, P= 0.19 for D. pulex) and (d) in mesocosms with no nutrient addition, low (Low) and high (High) addition (F2,43= 2.58, P= 0.09 for zooplankton, F2,43 = 2.07, P= 0.14 for D. pulex). Two chlorophyll a readings were negative and were not included in the analysis; the large box represents data between the 25th and 75th percentile; the line within the box represents the median. Vertical bars show the upper and lower 10 percentiles, dots represent outliers.

DISCUSSION

In a previous experiment using the same tanks, nutrients had a far greater impact on zooplankton, phytoplankton and macrophytes than temperature (McKee et al., 2002, 2003; Moss et al., 2003; Feuchtmayr et al., 2007). Our results, for a slightly greater temperature increase, suggest that temperature is an important driver of zooplankton growth and its algal food during spring. Our data allowed us: to quantify the extent to which the zooplankton peak abundance in spring is advanced by a 4°C warming, affected by nutrient loading; to test whether high nutrient loadings increase peak zooplankton abundance per unit of food; to test whether 4°C warming reduce peak abundance per unit of food as expected from increased energetic demands; and to explore the consequences for the aquatic system. Surprisingly, temperature and nutrient addition had contrasting impacts on the timing of phytoplankton and zooplankton peak abundances during this early stage of the growth season.

Increased temperature induced an earlier peak of chlorophyll a and zooplankton. As climate warming continues, the evidence from our experiment and studies so far (Gerten and Adrian, 2000; Scheffer et al., 2001; Berger et al., 2007) suggest an earlier zooplankton/Daphnia spring peak and subsequent clear water phase. We calculated an advance in zooplankton and Daphnia peak of 22–24 days with 4°C warming. The advance in peak zooplankton and Daphnia abundance in the heated tanks was consistent with findings from the Bautzen reservoir [23.2 days for 4°C warming (Wagner and Benndorf, 2007)] and Plußsee [20–24 days earlier for a 4°C increase (Müller-Navarra et al., 1997)], but was not as early as expected from the regression model of Scheffer et al. (2001) (33.6 days with 4°C warming) or from a linear extrapolation of findings from Lake Müggelsee [47 days for a 4°C increase (Adrian et al., 2006)]. The advance of the zooplankton and Daphnia peak with warming was larger than that of the phytoplankton peak, supporting findings of Schalau et al. (2008), a modelling study suggesting temperature and not food as the dominant factor for determining the timing of spring Daphnia populations. This phenomenon could be explained by the generally higher thermal sensitivity of growth and grazing of heterotrophs compared with primary producers (Allen et al., 2005; López-Urrutia et al., 2006; Rose and Caron, 2007; Bissinger et al., 2008).

With climate change, winter precipitation is expected to increase, leading to higher nutrient runoff from the catchment into rivers and lakes (IPCC, 2007), especially in the agricultural lowlands. With higher nutrient availability, phytoplankton growth is less likely to be limited by nitrogen or phosphorus which might result in high phytoplankton division rates leading to earlier peak abundances (Thackeray et al., 2008). However, phytoplankton abundance is also expected to increase with higher nutrient loading, and similar division rates could lead to higher and thus later peak abundances. Consequently, this might lead to higher and later peak grazer abundances. We found evidence for this hypothesis in our experiment. Chlorophyll a increased with nitrogen addition and the influence of nutrient addition on the date of Daphnia peak abundance was opposite to the temperature effect (Table II, Fig. 3); the higher the nutrient addition, the later the Daphnia peak abundance occurred. In contrast, model predictions of nutrient effects on transient Daphnia dynamics by Jäger et al. (2008) showed an advanced timing of the Daphnia peak biomass by phosphorus enrichment. Phosphorus additions used in this study and in the experiment by Jäger et al. (2008) were similar; however, nitrogen enrichment was not studied by Jäger et al. (2008).

Jäger et al. (2008) predicted and observed an increase in Daphnia peak biomass with phosphorus enrichment. A nutrient-enhanced increase in peak abundance of Daphnia is also consistent with the prediction of Schalau et al. (2008) who proposed that the height of this peak should be mostly controlled by phytoplankton carrying capacity, and hence may correlate with concentration of the limiting nutrient. Our findings confirmed the predicted increased mean chlorophyll a concentrations with nutrient addition (Table I), but the heights of the zooplankton and D. pulex peaks did not differ between nutrient treatments (Fig. 4), possibly due to crowding effects.

The abundance of Daphnia pulex in the mesocosms was unusually high. Crowding effects of Daphnia are reported to occur at densities of around 150 individuals per litre (Burns, 1995) in the laboratory. In the mesocosms, Daphnia pulex reached up to 10-fold higher abundances at peak times when food was plentiful. The simulation of Scheffer et al. (2001, their Figure 3) suggested a lower predicted peak of zooplankton abundance with 3°C warming, in accordance with the model of Müller-Navarra et al. (1997). We found higher zooplankton and Daphnia mean abundances with warming, but no difference in heights of the Daphnia abundance peaks with temperature. This lack of a thermal effect could be explained by the occurrence of Daphnia peak abundance at similar temperatures (mean ± SE of 15.3 ± 0.3°C for unheated and 16.6 ± 0.3°C for heated tanks) irrespective of thermal treatment (also see Gillooly and Dodson, 2000).

In these organic-rich hypertrophic systems, oxygen concentrations were reduced to critical levels for Daphnia survival in the heated tanks, due to respiration of open-water and sediment biota. Weider and Lampert (1985) experimentally identified threshold oxygen concentrations for Daphnia pulex of 0.5–1 mg L−1, below which survival of the animals decreased significantly. In 20 out of 23 heated mesocosms, mean oxygen concentrations at mid-depth around the zooplankton peaks were lower than 1 mg L−1, and Daphnia numbers were surprisingly high (see above). The contribution of macroinvertebrates to total respiration is assumed to be small compared with that of the zooplankton, phytoplankton and bacteria, as in this early stage of the mesocosm experiment, general macroinvertebrate abundances were still low (R.J. Moran et al., unpublished results). Increased total community respiration with warming along with decreasing phytoplankton is likely to have reduced the oxygen concentrations in heated mesocosms to levels that nonetheless probably limited Daphnia growth and reproduction.

Implications for freshwater systems of a temperature increase similar to that in our experiment are 3-fold. First, an increase of water temperature results in a lower dissolved oxygen saturation in shallow hypertrophic systems, and, together with high respiratory oxygen demands associated with both high heterotroph biomass and warming, can lead to severe oxygen depletion. This can affect not only survival of zooplankters but potentially also other invertebrates and fish (Moran et al., 2010). With climate warming, shallow eutrophic lakes might increasingly face problems resulting from very low dissolved oxygen concentrations. Secondly, while zooplankton can react rapidly to changes within their environment, the extent to which taxa from higher trophic levels, with longer generation times, such as invertebrate predators and fish, can adjust and cope with the earlier zooplankton peak needs to be addressed in further studies (match/mismatch hypothesis). As fish larvae and planktivorous fish species are dependent on zooplankton as a food source, climate change could have a detrimental effect on the food web in hypertrophic lakes, not only through very low oxygen concentrations but also because of an advanced timing of the zooplankton peak. Concluding from our results that effects of warming and high nutrient load on timing of peak abundances compensate each other might be misleadingly comforting. Even though nutrient addition delayed the peak abundances, mean concentration of chlorophyll a in High treatments was more than 2-fold higher than other treatments, and consequences of high nutrient loading to ecosystems are generally detrimental.

Thirdly, owing to increased temperatures and an earlier clear water phase, “non-grazeable” algae, such as cyanobacteria, might establish themselves earlier in the season and could then be able to dominate the community over the summer (Jeppesen et al., 2009). Results of the PROTECH model suggest that warming will increase phytoplankton biomass dominated by the cyanobacterium Anabaena during the summer bloom along with a loss of phytoplankton biodiversity (Elliott et al., 2006). Together with heavy nutrient loads, either already in the system or entering via runoff from the catchment, this could result in a substantial change of the ecosystem. However, through an early zooplankton peak and subsequent clear water phase in shallow lakes, the probability of macrophyte development might increase, decreasing the probability of a high phytoplankton biomass because of competition mechanisms.

So far, however, the general consensus is that establishment and stabilization of a macrophyte-dominated system requires a relative increase in the piscivorous fish stock and a reduction in nutrient levels (Gonzalez Sagrario et al., 2005). In our experimental ecosystems, despite the high nutrient levels, and in the absence of fish, macrophytes established themselves in all our mesocosms later that summer even in mesocosms in which planktivorous fish were introduced and became established for a subsequent experiment (Feuchtmayr et al., 2009). This indicates the establishment of a clear water state in our experimental mesocosms. However, the present study did not incorporate any influence of fish predation on zooplankton, including the effect of temperature on fish offspring and survival (Jeppesen et al., 2003) which might prevent such high zooplankton numbers from occurring and a clear water state from establishing.

Our findings that the timing of zooplankton peak abundance was advanced by 4°C warming and that oxygen was strongly depleted especially in heated tanks, together with evidence of differences in thermal sensitivity between autotrophic and heterotrophic rates (e.g. López-Urrutia et al., 2006; Rose and Caron, 2007) lead us to propose that an important component of future research on freshwater ecosystem responses to warming should include the difference in thermal sensitivities of rates of consumption, metabolism and production between autotrophs and heterotrophs. In contrast to current predictions of earlier peak plankton occurrence based on phosphorus additions, our finding of a delay in the timing of peak zooplankton and D. pulex abundances with phosphorus and nitrogen addition in hypertrophic systems suggests that nitrogen might be a more important driver in freshwater systems than currently accepted. Further, the strong influence of nutrient loading on timing and peak abundances reported here suggests studies on climate warming should include the analysis of nutrient effects.

FUNDING

Financial support was provided by the EU-grant Euro-limpacs (EU Contract No. GOCE-CT-2003-505540), funded under the EU Sixth Framework Programme Global Change and Ecosystems theme. Part of this work was conducted while D.A. was in receipt of a Sabbatical Fellowship at the National Center for Ecological Analysis and Synthesis (NCEAS) and latterly a Leverhulme Trust fellowship. NCEAS is funded by NSF (Grant no. DEB-0553768), the University of California, Santa Barbara and the State of California.

ACKNOWLEDGEMENTS

We are grateful to Jan Hatton, Ratcha Chaichana, Tom Heyes and Mike O'Connor for help with data collection and measurements, and S. Thackeray, S. Diehl and S. Maberly for helpful comments on an earlier draft of the manuscript. We thank S. Thackeray for support with statistical methods.

REFERENCES

Adrian
R.
Wilhelm
S.
Gerten
D.
Life-history traits of lake plankton species may govern their phenological response to climate warming
Global Change Biol.
 , 
2006
, vol. 
12
 (pg. 
652
-
661
)
Allen
A. P.
Gillooly
J. F.
Brown
J. H.
Linking the global carbon cycle to individual metabolism
Funct. Ecol.
 , 
2005
, vol. 
19
 (pg. 
202
-
213
)
Berger
S. A.
Diehl
S.
Stibor
H.
, et al.  . 
Water temperature and mixing depth affect timing and magnitude of events during spring succession of the plankton
Oecologia
 , 
2007
, vol. 
150
 (pg. 
643
-
654
)
Biggs
J.
Williams
P.
Whitfield
M.
, et al.  . 
15 years of pond assessment in Britain: results and lessons learned from the work of Pond Conservation
Aquat. Conserv. Mar. Freshwater Ecosyst.
 , 
2005
, vol. 
15
 (pg. 
693
-
714
)
Bissinger
J. E.
Montagnes
D. J. S.
Sharples
J.
, et al.  . 
Predicting marine phytoplankton maximum growth rates from temperature: improving on the Eppley curve using quantile regression
Limnol. Oceanogr.
 , 
2008
, vol. 
53
 (pg. 
487
-
493
)
Burns
C. W.
Effects of crowding and different food levels on growth and reproductive investment of Daphnia
Oecologia
 , 
1995
, vol. 
101
 (pg. 
234
-
244
)
Carvalho
L.
Kirika
A.
Changes in shallow lake functioning: response to climate change and nutrient reduction
Hydrobiologia
 , 
2003
, vol. 
506
 (pg. 
789
-
796
)
Diehl
S.
Berger
S.
Ptacnik
R.
, et al.  . 
Phytoplankton, light, and nutrients in a gradient of mixing depths: field experiments
Ecology
 , 
2002
, vol. 
83
 (pg. 
399
-
411
)
Elliott
J. A.
Jones
I. D.
Thackeray
S. J.
Testing the sensitivity of phytoplankton communities to changes in water temperature and nutrient load, in a temperate lake
Hydrobiologia
 , 
2006
, vol. 
559
 (pg. 
401
-
411
)
Feuchtmayr
H.
McKee
D.
Harvey
I. F.
, et al.  . 
Response of macroinvertebrates to warming, nutrient addition and predation in large-scale mesocosm tanks
Hydrobiologia
 , 
2007
, vol. 
584
 (pg. 
425
-
432
)
Feuchtmayr
H.
Moran
R.
Hatton
K.
, et al.  . 
Global warming and eutrophication: effects on water chemistry and autotrophic communities in experimental hypertrophic shallow lake mesocosms
J. Appl. Ecol.
 , 
2009
, vol. 
46
 (pg. 
713
-
723
)
Gerten
D.
Adrian
R.
Climate-driven changes in spring plankton dynamics and the sensitivity of shallow polymictic lakes to the North Atlantic Oscillation
Limnol. Oceanogr.
 , 
2000
, vol. 
45
 (pg. 
1058
-
1066
)
Gillooly
J. F.
Dodson
S. I.
Latitudinal patterns in the size distribution and seasonal dynamics of new world, freshwater cladocerans
Limnol. Oceanogr.
 , 
2000
, vol. 
45
 (pg. 
22
-
30
)
Gonzalez Sagrario
M. A.
Jeppesen
E.
Goma
J.
, et al.  . 
Does high nitrogen loading prevent clear-water conditions in shallow lakes at moderately high phosphorus concentrations?
Freshwater Biol.
 , 
2005
, vol. 
50
 (pg. 
27
-
41
)
Solomon
S.
Qin
D.
Manning
M.
Chen
Z.
Marquis
M.
Averyt
K. B.
Tignor
M.
Miller
H. L.
IPCC.
Summary for Policymakers
 , 
2007
Cambridge
Cambridge University Press
 
UK and New York, NY, USA
Jäger
C. G.
Diehl
S.
Matauschek
C.
, et al.  . 
Transient dynamics of pelagic producer grazer systems in a gradient of nutrients and mixing depths
Ecology
 , 
2008
, vol. 
89
 (pg. 
1272
-
1286
)
Jeppesen
E.
Sondergaard
M.
Jensen
J. P.
Climatic warming and regime shifts in lake food webs—some comments
Limnol. Oceanogr.
 , 
2003
, vol. 
48
 (pg. 
1346
-
1349
)
Jeppesen
E.
Kronvang
B.
Meerhoff
M.
, et al.  . 
Climate change effects on runoff, catchment phosphorus loading and lake ecological state, and potential adaptations
J. Environ. Qual.
 , 
2009
, vol. 
38
 (pg. 
1930
-
1941
)
Johnes
P. J.
Heathwaite
A. L.
A procedure for the simultaneous determination of total nitrogen and total phosphorus in fresh-water samples using persulfate microwave digestion
Water Res.
 , 
1992
, vol. 
26
 (pg. 
1281
-
1287
)
Johnk
K. D.
Huisman
J.
Sharples
J.
, et al.  . 
Summer heatwaves promote blooms of harmful cyanobacteria
Global Change Biol.
 , 
2008
, vol. 
14
 (pg. 
495
-
512
)
López-Urrutia
A.
San Martin
E.
Harris
R. P.
, et al.  . 
Scaling the metabolic balance of the oceans
Proc. Natl Acad. Sci. USA
 , 
2006
, vol. 
103
 (pg. 
8739
-
8744
)
Mackereth
F. J. H.
Heron
J.
Talling
J. F.
Water analysis: Some revised methods for limnologists
Freshwater Biol. Assoc. Publ
 , 
1989
, vol. 
36
 pg. 
120
 
McKee
D.
Atkinson
D.
Collings
S. E.
, et al.  . 
Heated aquatic microcosms for climate change experiments
Freshwater Forum
 , 
2000
, vol. 
14
 (pg. 
51
-
58
)
McKee
D.
Hatton
K.
Eaton
J. W.
, et al.  . 
Effects of simulated climate warming on macrophytes in freshwater microcosm communities
Aquat. Bot.
 , 
2002
, vol. 
74
 (pg. 
71
-
83
)
McKee
D.
Atkinson
D.
Collings
S. E.
, et al.  . 
Response of freshwater microcosm communities to nutrients, fish, and elevated temperature during winter and summer
Limnol. Oceanogr.
 , 
2003
, vol. 
48
 (pg. 
707
-
722
)
Mooij
W. M.
Hulsmann
S.
Domis
L. N. D.
, et al.  . 
The impact of climate change on lakes in the Netherlands: a review
Aquat. Ecol.
 , 
2005
, vol. 
39
 (pg. 
381
-
400
)
Mooij
W. M.
Janse
J. H.
Domis
L.
, et al.  . 
Predicting the effect of climate change on temperate shallow lakes with the ecosystem model PCLake
Hydrobiologia
 , 
2007
, vol. 
584
 (pg. 
443
-
454
)
Moran
R.
Harvey
I.
Moss
B.
, et al.  . 
Influence of simulated climate change and eutrophication on three-spined stickleback populations: a large scale mesocosm experiment
Freshwater Biol.
 , 
2010
, vol. 
55
 (pg. 
315
-
325
)
Moss
B.
McKee
D.
Atkinson
D.
, et al.  . 
How important is climate? Effects of warming, nutrient addition and fish on phytoplankton in shallow lake microcosms
J. Appl. Ecol.
 , 
2003
, vol. 
40
 (pg. 
782
-
792
)
Müller-Navarra
D. C.
Güss
S.
Von Storch
H.
Interannual variability of seasonal succession events in a temperate lake and its relation to temperature variability
Global Change Biol.
 , 
1997
, vol. 
3
 (pg. 
429
-
438
)
Peeters
F.
Straile
D.
Lorke
A.
, et al.  . 
Earlier onset of the spring phytoplankton bloom in lakes of the temperate zone in a warmer climate
Global Change Biol.
 , 
2007
, vol. 
13
 (pg. 
1898
-
1909
)
Rolinski
S.
Horn
H.
Petzoldt
T.
, et al.  . 
Identifying cardinal dates in phytoplankton time series to enable the analysis of long-term trends
Oecologia
 , 
2007
, vol. 
153
 (pg. 
997
-
1008
)
Rose
J. M.
Caron
D. A.
Does low temperature constrain the growth rates of heterotrophic protists? Evidence and implications for algal blooms in cold waters
Limnol. Oceanogr.
 , 
2007
, vol. 
52
 (pg. 
886
-
895
)
Savage
V. M.
Gillooly
J. F.
Brown
J. H.
, et al.  . 
Effects of body size and temperature on population growth
Am. Nat.
 , 
2004
, vol. 
163
 (pg. 
429
-
441
)
Schalau
K.
Rinke
K.
Straile
D.
, et al.  . 
Temperature is the key factor explaining interannual variability of Daphnia development in spring: a modelling study
Oecologia
 , 
2008
, vol. 
157
 (pg. 
531
-
543
)
Scheffer
M.
Hosper
S. H.
Meijer
M. L.
, et al.  . 
Alternative equilibria in shallow lakes
Trends Ecol. Evolut.
 , 
1993
, vol. 
8
 (pg. 
275
-
279
)
Scheffer
M.
Straile
D.
van Nes
E. H.
, et al.  . 
Climatic warming causes regime shifts in lake food webs
Limnol. Oceanogr.
 , 
2001
, vol. 
46
 (pg. 
1780
-
1783
)
R Development Core Team
R: A Language and Environment for Statistical Computing
R Foundation for Statistical Computing
 , 
2007
Vienna, Austria
Thackeray
S. J.
Jones
I. D.
Maberly
S. C.
Long-term change in the phenology of spring phytoplankton: species-specific responses to nutrient enrichment and climatic change
J. Ecol.
 , 
2008
, vol. 
96
 (pg. 
523
-
535
)
Van Donk
E.
Santamaria
L.
Mooij
W. M.
Climate warming causes regime shifts in lake food webs: a reassessment
Limnol. Oceanogr.
 , 
2003
, vol. 
48
 (pg. 
1350
-
1353
)
Wagner
A.
Benndorf
J.
Climate-driven warming during spring destabilises a Daphnia population: a mechanistic food web approach
Oecologia
 , 
2007
, vol. 
151
 (pg. 
351
-
364
)
Weider
L. J.
Lampert
W.
Differential response of Daphnia genotypes to oxygen stress: respiration rates, hemoglobin content and low-oxygen tolerance
Oecologia
 , 
1985
, vol. 
65
 (pg. 
487
-
491
)
Werner
E. E.
Skelly
D. K.
Relyea
R. A.
, et al.  . 
Amphibian species richness across environmental gradients
Oikos
 , 
2007
, vol. 
116
 (pg. 
1697
-
1712
)
Weyhenmeyer
G. A.
Warmer winters: are planktonic algal populations in Sweden's largest lakes affected?
Ambio
 , 
2001
, vol. 
30
 (pg. 
565
-
571
)
Weyhenmeyer
G. A.
Blenckner
T.
Pettersson
K.
Changes of the plankton spring outburst related to the North Atlantic Oscillation
Limnol. Oceanogr.
 , 
1999
, vol. 
44
 (pg. 
1788
-
1792
)
Winder
M.
Schindler
D. E.
Climatic effects on the phenology of lake processes
Global Change Biol.
 , 
2004
, vol. 
10
 (pg. 
1844
-
1856
)

Author notes

Corresponding editor: Beatrix E. Beisner