Short term fluctuating temperature alleviates Daphnia stoichiometric constraints

In this study, we analysed how short term temperature fluctuation interacts with nutrient limitation in the vertical migrating Daphnia commutata. We hypothesize that short term (daily) temperature fluctuation will alleviate nutrient limitation. We carried out experiments analysing growth rates, phosphorus and RNA content of D. commutate grown under four different temperature regimes and two P-limited conditions. Our experiments showed that individuals grown under fluctuating temperature grew more than at the mean temperature. We estimated the expected sizes for the 15 °C treatment based on the Q10 and for the fluctuating temperature treatment. These expected sizes for both treatments resulted well below the observed ones. The P and RNA content of individuals grown at 10 °C were significantly higher than those at 20 °C, and when individuals grown at 10 °C were translocated to 20 °C they exerted an increased growth rate. Our results suggest that, under a regime of diel vertical migration, the temperature alternation would allow migrating organisms to alleviate the effect of severe nutrient limitation maintaining population growth. Under a scenario of global warming, where epilimnetic temperatures will increase, lake temperature will interact with nutrient limitation for consumers, but, organisms may be able to face these changes if they can still regularly move from a cold hypolimnion to a warmer epilimnion.


Results
Experiment 1. Growth rates, P and RNA content were contrasted between D. commutata grown at constant temperatures (10 °C or 20 °C) for 7 days with those grown at low temperature for 2 days and then translocated to warm temperature for the remaining 5 days. All neonates at the start of the experiment were of similar size (2.81 ± 0.15 µg ind −1 ) (ANOVA F 2,24 = 3.14, p = 0.061). Neonates in the cold treatment reached, after 7 days, a mean size 5.66 µg ind −1 , smaller than individuals of all other treatments. Mean size of individuals in the 20 °C treatment after 7 days was 21.58 µg ind −1 . As expected, the size of individuals maintained at 10 °C for 2 days was equal to those of the cold treatment at the same age, but when translocated to 20 °C reached 17.62 µg ind −1 in the next 5 days. There were significant differences in size between all treatments with a significant interaction (twoway RM ANOVA time x treatment F 6,72 = 54.51, p < 0.001). As all neonates were of similar size at the beginning, this interaction indicates that differences among treatments increased over time.
After the first 2 days, we measured the C:P of Daphnia from the treatments at 20 °C (warm treatment) and at 10 °C (cold and cold-warm treatments). The results showed that the C:P of the individuals at 10 °C was significantly lower than at 20 °C (t-test t (6) = 2.78, p = 0.031) (Fig. 2), indicating a higher P content in the individuals grown at 10 °C. Indeed, the P content (%P to DW) of individuals at 10 °C was 2.32%, while the P content of those at 20 °C was 1.2%. Individuals in the cold treatment had significantly higher RNA content after 2 days than individuals of the same age at 20 °C (3.6% and 1.6% respectively), and also higher than both treatments at day 4 (all at 20 °C) (one-way ANOVA F 3,28 = 12.99, p < 0.001) (Fig. 3).

Experiment 2.
We compared four temperature treatments (cold, medium, warm and fluctuating) under P limiting condition (C:P = 450). As expected, organisms grew faster in warmer temperatures than in the cold treatment. Even though the mean temperature for both medium and fluctuating temperature treatments was very close (15 and 14.88 °C respectively), organisms in the fluctuating treatment grew faster than in the medium temperature, and even faster than the linear interpolation of the growth at 10 and 20 °C (Fig. 4a). There was a significant interaction between time and temperature (two-way RM ANOVA time x treatment F 15     www.nature.com/scientificreports/ RNA analysis in the fluctuating treatment (at 60 h, end of cold period 3.2%, and 72 h, end of warm period 3.9%), did not show significant differences (t-test t (14) = 0.97, p = 0.347).
We calculated the expected sizes of the 15 °C and fluctuating treatments. The results obtained showed that the expected sizes were well below the observed ones ( Fig. 5 solid versus dashed lines), with all expected values, except for the first two or three days, laying beyond the 95% confidence limits (Fig. 5). Noticeably, the expected and observed sizes of the fluctuating temperature were larger than the corresponding ones of the 15 °C treatment ( Fig. 5 grey lines compared to black lines).

Experiment 3.
In this experiment we compared D. commutata growing at 15 °C with the ones in the fluctuating temperature, but with a higher P limitation than in experiment 2 (C:P = 650). We observed the same trend as in experiment 2, that is the individuals in the fluctuating temperature treatment grew more than at constant 15 °C (Fig. 4b). As expected, animals in experiment 3 fed with lower food quality, grew less than those of experiment 2. However, the decrease in size of Daphnia in experiment 3 (lower food quality, C:P = 650) compared to experiment 2 (C:P = 450), was smaller in the fluctuating treatment (21%) compared to the constant 15 °C (26%), and these differences were significant (two way ANOVA Temp F 1,36 , p = 0.005; C:P F 1,36 p = 0.004), and in the a posteriori Tukey test, all comparisons were significant but C:P within fluctuating (p = 0.065).

Discussion
In this study we show that under fluctuating temperature regimes, as those imposed to migrating zooplankton, organisms can growth more than predicted by temperature alone. In our experiments, in the fluctuating temperature regime, organisms reached sizes not only larger than expected by the Q 10 , but even larger than the mean size between the 10 °C and 20 °C treatments, assuming a linear interpolation as a conservative supposition. However, metabolic functions including growth do not increase linearly with temperature 1,43 . According to the  www.nature.com/scientificreports/ exponential response of metabolic and growth rates to temperature and the Q 10 estimations, it would be expected that the growth at mean temperature (15 °C) should be lower than the linear interpolation of two temperatures (10 and 20 °C). In addition, the sizes of the 15 °C treatment were also larger than expected by Q 10 . These larger than expected sizes, would imply that other factor than temperature, is also affecting growth.
Our experiments also showed that D. commutata grown at low temperature had significantly more P and RNA content than individuals grown at higher temperature. In addition, individuals raised at low temperature (10 °C) also responded to increasing temperature (20 °C) with increased growth. Woods, et al. 16 found that there is a general trend of P and RNA increase at low temperature, probably because more rRNA is needed to maintain synthesis at low temperatures 18 . The efficiency of synthesis, the synthesis per rRNA unit, decreases with the decrease in temperature so an increase in rRNA, and hence in P, may compensate at least in part for the decrease in synthesis 17 . Our results on RNA content gave rather low values of this compound, however, low RNA content were also obtained in experiments with Daphnia galeata and D. pulicharia under similar P-limited conditions (C:P of food 457) 44 . In our first experiment we observed that growth rate and RNA content showed opposite patterns. High RNA content was registered when GR was low (0-2 days at 10 °C) while high GR was observed with low RNA content (0-2 days at 20 °C and 2-4 days in the 10-20 °C treatment). A decoupling of the RNA-GR relationship was observed when the GRH was tested across temperatures, and was attributed to a less RNA requirement at high temperatures due to faster synthesis 7 . In our first experiment, during the first two days, GR in the 10 °C treatment was low, and in the next period when translocated to 20 °C (2-4 days) it was high. The GR is calculated for a certain period of time, while the RNA is a measurement of a specific moment, that does not integrate the period as GR does. Consequently, the GR of the 2-4 period does not depend on the RNA at the end of the period, but at the beginning of it, which was high. As growth takes place, the increase in size produces a dilution effect if DW increases more than P uptake.
Whether P limitation increases or decreases with temperature is still under discussion 15 . In any case, individuals grown at lower temperature have a higher P content, and this condition is achieved in 48 h, showing that is not necessarily an adaptive condition of populations living permanently at different temperatures 4 . It can be assumed that individuals at low temperature would incorporate P from the diet at a similar rate than individuals at higher temperature, but their growth will be lower, so the P and RNA content is not diluted with the increase in body mass. Alternatively, at 20 °C the body grows faster, so the P and RNA content (µg RNA µg DW −1 ) is lower because of a dilution effect with higher DW. Indeed, in our first experiment, after 48 h individuals grown at 10 °C and 20 °C had similar P and RNA per individual, but very different P and RNA content due to the larger size of individuals grown at 20 °C. Consequently, when individuals growing at 10 °C for 2 days were transferred to 20 °C, the higher RNA content (%RNA), allowed them to increase their growth rate reaching even higher growth rates than individuals growing permanently at 20 °C in the 0-2 days. That is, when individuals at 10 °C were changed to 20 °C, they were less P limited and the combined effect of high temperature and high P and RNA content, allowed a faster growth.
Although the activity of alkaline phosphatase enzyme (APA) has been considered as a proxy of P limitation, it has been shown that at low temperatures, APA increases to compensate for the lower enzymatic efficiency 10 . These measurements were carried out after 6 days of food and temperature acclimation, but we don't have information on how quick APA can change. Some defense enzymes, as those required to neutralize reactive oxygen species that induce oxidative stress, can react very quickly (1 h) 45 , while APA changes were observed after 4 h of induced oxidative stress (ultraviolet light or H 2 O 2 exposure) 46 . Because of its function, that is very different from that of antioxidant enzymes, it is unlikely that APA would respond so fast. This means that during regular short time temperature fluctuation as in DVM (mimicked in our experiment) APA would be not as high as the low temperature would induce, but much higher than that of 20 °C. If APA does not respond quickly, when individuals migrate upwards during dusk, they will be at higher temperature (fast growing) with much more APA than individuals permanently at 20 °C, so they could uptake P more efficiently. Probably, this condition of high APA can be extended for few hours, giving the organism a better somatic condition for growth. Then, they return to colder waters, and the growth its reduced by the low temperature, but some hours later, APA would be again increased. This effect would explain the higher growth rate observed in the fluctuating temperature treatment (higher than expected only by temperature) and also the lack of differences in P and RNA content in the different phases of the temperature cycle.
The increased growth that occurs following a period of restriction (mainly food) was described as compensatory growth 47,48 . However, copepods that were fed with P-deficient food did not show this response in growth 49 . In our experiment we did not vary food P content or food quantity, but temperature, and we did observe an increased growth when the temperature was risen to 20 °C. This "thermal compensatory" growth lasts for a short time, probably while %RNA remains high, but as the individuals grow quickly, the %RNA decreases along with the increase in size, as seen in days 2-4 and 4-7 of experiment 1.
In the fluctuating temperature treatments, we expected that at the end of the 12 h cold period, D. commutata would have more RNA content than at the end of the warm period (12 h later), similar to the results obtained in experiment 1 after 48 h. However, in the fluctuating temperature regime, the RNA content was similar to the cold treatment of experiment 1, so what we did not see was a significant decrease during the warm period, remaining high during both phases of the temperature cycle. In addition to the temperature effect on APA, discussed previously, it could be assumed that the cold period is like a nutrient pulse in the sense of the Droop model 50 , so the increase in P content in D. commutata would have occurred in a very short time period, as it occurs in algae when a P-spike is added 18,51,52 . However, during the cold period P content would increase much more slowly than expected by the Droop model with a P pulse. Phosphorus uptake may not change during the cold periods, but because of the slow growth, relative P content increases. The result is that under this fluctuating regime, individuals spending the 20 °C period with the RNA content of the 10 °C period can grow faster than expected by temperature alone. www.nature.com/scientificreports/ Daily migration to cold water was seen as a behaviour that lowers growth, but with the advantage of avoiding visual predation 23,39 or damaging wavelengths 22 . However, here we show that under nutrient-limitation, daily migration can have other advantages. The time spent in the cold water will be in part compensated by increased growth while in warm water, due to the P uptake during the cold period. It has been argued that during DVM, the organisms fast while in the cold deep water, since the food was only in the warm illuminated epilimnion 21,27 . However, the presence of a deep Chlorophyll maxima is frequent in many lakes around the world [53][54][55] . Moreover, the depth of the deep chlorophyll maxima depends primarily on light more than on other factors 33,35 , and in transparent systems, this deep maxima lays below the thermocline, so cold water layers are not food depleted. In addition, the C:P ratio of the algae would be lower (better food quality for grazers) at the hypolimnion 32,36 where the DCM lays. This implies that migrating individuals not only do not face fasting while in the deep cold water as previously suggested 27 , but may have more and better food. In our experiments, we did not vary food quantity or quality during changes in temperature (simulating DVM), however, we showed that fluctuating temperature alone increases growth. In addition, the effect of increased food quality in deep water 32,36 would imply that grazers would have more P and the effect we reported here would be enhanced. Under a DVM regime, the temperature alternation with the concomitant nutrient acquisition in cold water, would allow migrating organisms to alleviate the effect of severe nutrient limitation maintaining population growth. Thus, the costs of vertical migration, in terms of reduced growth rate 27 , may be much lower than determined in early experiments in which Daphnia were fed with P-replete algal food. Global warming is affecting the temperature structure of lakes, increasing the stratification period, but also increasing the thermal difference between epilimnion and hypolimnion 56,57 . These changes in lake temperature will interact with nutrient limitation for consumers, but, according to our results, organisms may be able to face these changes if they can still regularly move from a cold hypolimnion to a warmer epilimnion.

Methods
All experiments utilised a clone culture of Daphnia commutata originally started from a female collected from Lake Mascardi (41°15′-41°25′S; 71°28′-71°39′ W) in Patagonia, Argentina, and maintained in the laboratory for more than ten years in P-free COMBO medium modified from Kilham, et al. 58 , fed Chlamydomonas reinhardtii in a chamber at 15 °C. The C. reinhardtii culture was grown in MBL medium 59 . The P source (K 2 HPO 4 ) of the algal culture was controlled in each experiment as detailed below in the experimental design. All experiments were conducted in water baths in growth chambers at the selected temperature and subject to a photoperiod of 12:12 (light:dark) with dim light (4 µmol photon m −2 s −1 ). We used this light level for the illuminated phase of the experimental day cycle because during the day D. commutata migrating individuals are located in the hypolimnetic deep cold water, around the 1% surface irradiance 41,42 . All experiments began using neonates (< 12 h old) from adult females fed with algae grown in P-limited media according to each experiment. Experimental design. Experiment 1. Neonate (< 12 h old) D. commutata were placed individually in 15 mL tubes with P-free COMBO and P-limited but not quantity limited food (Chamydomonas reinhardtii: 1 mg C L −1 , C:P atomic ratio of 450). Previous experiments showed that at a C:P of 450 D. commutata were P limited and it is a food C:P ratio commonly encountered by D. commutata in its natural environment 36,41,60 . Medium and food were replaced every day by pipetting the individual to a clean tube with fresh medium and food. Adults of this species (2 mm in body length) have a clearance rate of 4 mL d −161 . Assuming a clearance rate of 4 mL d −1 , much higher than for juveniles, at the end of the 24 h period, the C concentration would have reduced to 0.75 mg C L −1 , thus, in our experiments, neonates and juveniles, that reached 1.6 mm in body length, were not food limited between food replacements. The treatments were (a) cold (b) cold-warm and (c) warm (Fig. S1). Briefly, 80 individuals were placed at 10 °C and 65 individuals at 20 °C. After two days, individuals from the cold treatment (10 °C) were randomly distributed as follows: 15 were kept at 10 °C for cold water growth estimations, 25 were switched to 20 °C (cold-warm treatment), 30 used for C and P determinations (3 replicates of 5 individuals for P and equal set for C) and 10 for RNA determination. On the same day, of the 65 individuals at 20 °C (warm treatment), 30 were designated for C and P content, and 10 for RNA content. On day 4, 10 individuals from the cold-warm and warm treatments were designated for RNA determination and the remaining 15 each of coldwarm and warm treatments were kept at 20 °C for growth estimations until the end of the experiment. Growth estimations were based on size measurements of each individual in each treatment at the initial time (0 day), 2 days, 4 days and 7 days as detailed in the measurements section. Experiment 2. D. commutata was subjected to four treatments (10 °C, 15 °C, 20 °C and fluctuating temperature regime). In all treatments, individuals were placed individually in 15 mL tubes, with P-free COMBO and food (C. reinhardtii 1 mg C L −1 ) with a C:P atomic ratio of 450. Medium and food were replaced every day by pipetting the individual to a clean tube with fresh medium and food. The treatments were (a) cold: 15 individuals were placed at 10 °C for 9 days; (b) medium: 15 individuals were placed at 15 °C for 9 days; (c) warm: 15 individuals were placed at 20 °C for 9 days; and (d) fluctuating regime: 35 individuals were placed for 9 days in a system that changed the temperature in a continuous cycle of 12 h at 9 °C and 12 h at 20 °C, with a dim light (4 µmol photon m −2 s −1 ) during the cold period and darkness during the warm period. This treatment was started with 35 individuals, and at 60 h (end of a cold period) and 72 h (end of a warm period), 10 individuals were sampled for RNA measurement. Thus, 15 individuals were used for size and growth, and 20 for RNA analysis. The rate of change between the two temperatures was 0.14 °C min −1 , requiring 75 min to reach the new temperature (Fig. S2). Sharp changes in temperature exert a stress effect 24 . In fact, Mikulski, et al. 24 found that growth rates and heat shock proteins dropped when the temperature changed at rates higher than ~ 0. 4  www.nature.com/scientificreports/ in our experiments, the rate of temperature change (0.14 °C min −1 ) was much lower, allowing us to assume no stress was experienced at this temperature change rate. The temperature was monitored with a data logger (HOBO Pendant Temp UA-001-08) that allowed for determination of possible temperature cycle anomalies during the experiment. The mean temperature of the water bath for the nine days of the experiment was 14.88 °C. This treatment was designed to mimic the temperature change during DVM, assuming equal time spent in epilimnion and hypolimnion. Individuals from all treatments were photographed every day. At the end of the warm period of the fluctuating treatment, the media and food were replaced. Experiment 3. Similar to experiment 2, we ran a third experiment with 2 treatments (15 °C and fluctuating temperature) with identical conditions but with a higher P limitation (C:P = 650). This condition of high C:P ratio was observed to be limiting for D. commutata growth 36 , and distribution 41 . In this case, we used 15 individuals for each treatment, as only size was measured in this experiment. This experiment was carried out to detect if the fluctuating temperature had similar effects compared with the mean temperature when individuals were more P limited.
As expected for this species in these food qualities, in all the three experiments D. commutata females did not develop eggs during the growth phase studied (9 days).
Size and growth estimations. Size and growth were estimated according to 44 using lateral images that allow measurement of individuals while the experiment is ongoing. Lateral images were obtained at 20 × magnification (Olympus SZX9 microscope) with a digital camera (5Mp). Images were analysed with ImageJ software (ImageJ, ver. 1.52a, NIH, USA). Biomass was estimated from area-dry weight regressions estimated for this species grown under similar food qualities 60 . Growth rates were estimated for each interval (0-2, 2-4 and 4-7) of experiment 1 and daily for experiments 2 and 3 as GR = lnDW f −lnDW i t in which DW f and DW i are the dry weights at the end and the beginning of the growth interval, respectively, and t is the time in days between measurements. Growth rates of experiment 2 were used for calculation of the Q 10 (see "Expected sizes and growth rates" section).
Carbon concentration of the food (all experiments) and of D. commutata from experiment 1 was estimated as follows. Algae: 3 mL of culture were filtered on GF/F precombusted filters and analysed in a Thermo Finnigan EA1112 elemental analyser. For D. commutata, 5 individuals were pooled on a GF/F precombusted filter and analysed as the algae. Phosphorus was analysed on equal volume of algae culture or Daphnia number in MilliQ water and combusted with Sodium persulfate at 125 °C and analysed with the ascorbate-reduced molybdenum method 62 .
Methods in Gorokhova and Kyle 63 and Hessen, et al. 18 were followed for RNA analyses: individuals were collected with forceps and gently rinsed in MilliQ RNAse-free water, placed alive individually in nuclease-free Eppendorf tubes, immediately snap-frozen in liquid N and then stored at − 80 °C until RNA analysis. Before the analysis, 100 µL of sarcosyl 1% in TE buffer (Tris-EDTA buffer, pH 7.5) was added to the tube, the tube kept on ice and sonicated for 90 s (Labsonic M, Sartorius) then diluted (1:5) with TE buffer. The tubes were then centrifuged (12,000 RPM at 4 °C) and the supernatant was used for analysis. For each individual 100 µL extracted supernatant was placed in a 96-well microplate and a second 100 µL in a contiguous well. To the latter, we added 1 µL of RNAse and incubated the plate in the dark at 37 °C for 1 h. Finally, 100 µL of Ribogreen was added to each well and the fluorescence read in a plate reader (Synergy HTX, BioTek) with excitation at 480 nm and emission at 520 nm. A calibration curve was placed in the first two lines of the plate, and the estimation was carried out as the difference in fluorescence between untreated and RNAse-treated wells. RNA content was expressed as percent of the RNA on dry weight (RNA/DW*100).
Expected sizes and growth rates. We estimated the expected sizes at 15 °C ( S (15)t ) based on the estimation of Q 10 , calculated on 10 °C and 20 °C treatments of experiment 2, Q 10 was calculated as where GR 1 and GR 2 are growth rates at temperatures T 1 and T 2 (T 2 > T 1 ).
Based on the Q 10 and the GR 10 we estimated the expected growth rates at 15 °C ( GR 15 ) as Then, we estimated the expected sizes at 15 °C ( S (15)t ) as We also estimated the expected sizes of the fluctuating temperature treatment ( S (f )t ) as successive cycles of alternating half day growing at a rate GR 10 (10 °C) and half at a rate GR 20 (20 °C). Although the daily growth rates (between consecutive days) were very similar to the mean growth rate (0-9 days), in this case we used for each day the observed GR for that period.

Statistical analyses.
Comparison of sizes at the beginning of the three experiments was conducted with a one-way ANOVA. Size change with time in each experiment and growth rates of experiment 1 were ana-