Species interactions mediate thermal evolution

Abstract Understanding whether populations and communities can evolve fast enough to keep up with ongoing climate change is one of the most pressing issues in biology today. A growing number of studies have documented rapid evolutionary responses to warming, suggesting that populations may be able to persist despite temperature increases. The challenge now is to better understand how species interactions, which are ubiquitous in nature, mediate these population responses to warming. Here, we use laboratory natural selection experiments in a freshwater community to test hypotheses related to how thermal evolution of Daphnia pulex to two selection temperatures (12 and 18°C) is mediated by rapid thermal evolution of its algal resource (Scenedesmus obliquus) or by the presence of the zooplankton predator Chaoborus americanus. We found that cold‐evolved algae (a high‐quality resource) facilitated the evolution of increased thermal plasticity in Daphnia populations selected at 12°C, for both body size and per capita growth rates (r). Conversely, warm‐evolved algae facilitated the evolution of increased r thermal plasticity for Daphnia selected at 18°C. Lastly, we found that the effect of selection temperature on evolved Daphnia body size was more pronounced when Daphnia were also reared with predators. These data demonstrate that trait evolution of a focal population to the thermal environment can be affected by both bottom‐up and top‐down species interactions and that rapid temperature evolution of a resource can have cascading effects on consumer thermal evolution. Our study highlights the importance of incorporating species interactions when estimating ecological and evolutionary responses of populations and communities to ongoing temperature warming.

2018; Tseng & O'Connor, 2015). These and many other ecological and evolutionary studies of temperature responses have collectively demonstrated that temperature affects all levels of organization in the natural world and that evolutionary responses of focal populations to the thermal environment can occur at ecological time scales.
Bottom-up species interactions, such as those between resources and their consumers, can affect thermal responses of consumers if the resources themselves are rapidly evolving to temperature. For example, resources that occupy the lower rungs of aquatic and terrestrial food chains (e.g., phytoplankton, microbes and insects) are also the organisms mostly likely to evolve rapidly, and these trait changes may have important consequences for thermal evolution of higher trophic levels. Of particular global importance are the trophic consequences of phytoplankton thermal evolution. These microscopic organisms fix over 50% of the planet's carbon dioxide, they produce over 50% of the Earth's oxygen, and they are the primary source of nutrients such as proteins and essential fatty acids, which cascade up the aquatic food web and that are ultimately consumed by humans (Finkel et al., 2010).
Conversely, top-down species interactions, such as those between predators and prey (for consistency, here we will refer to the prey as consumers), can also affect thermal evolution if predators depress the consumer population below which evolutionary responses are possible, or if predators preferentially eat consumers that happen to be particularly bad, or good, at withstanding the change in temperature (Osmond et al., 2017). Given the speed with which climate is warming around the world, there is a pressing need to better understand how these common species interactions affect thermal evolution. Here, we use a cosmopolitan freshwater food web to investigate the role of bottom-up (phytoplankton thermal evolution) and top-down (presence of a voracious predator) species interactions on rapid thermal evolution in the zooplankton Daphnia pulex (Cladocera: Daphniidae, Leydig 1860).
Our tri-trophic study system consists of the phytoplankton Scenedesmus obliquus (Chlorococcales: Scenedesmaceae, Kützing, 1833, also known as Tetradesmus obliquus and Acutodesmus obliquus), Daphnia pulex, and larvae of the predatory midge Chaoborus americanus (Diptera: Chaoboridae; De Haan 1849). We use controlled laboratory experiments to assess the effects of a) S. obliquus thermal evolution, and b) predator presence, on Daphnia thermal reaction norm (TRN) evolution in response to two temperature selection treatments (12 and 18°C).

| Predictions
Given their rapid doubling time (1-2 days) and large population size during the experiment (~500,000), we expected S. obliquus to show rapid thermal evolution.
This temperature evolution manifests as shifts in cell size, whereby cell size decreases as temperature increases (Chen, Li, Dai, Sun, & Chen, 2011;Margalef, 1954). Scenedesmus obliquus also produce fewer essential fatty acids at warmer temperatures, and warmreared S. obliquus sustain smaller populations of Daphnia compared to cold-reared algae (Breuer, Lamers, Martens, Draaisma, & Wijffels, 2013;Sikora, Dawidowicz, & Elert, 2014;Vigeolas et al., 2012). We thus expect warm-evolved S. obliquus to be a lower-quality food resource for Daphnia, and we propose two hypotheses for how phytoplankton thermal evolution may affect ecological and evolutionary responses of Daphnia to temperature-mediated selection (Box , Hypotheses 1-2). Cold-evolved (higher quality) algae could facilitate Daphnia thermal reaction norm (TRN) evolution if this food type sustains higher populations of Daphnia, thus maintaining the standing genetic variation on which thermal selection can act (Frankham, 1996). Concomitantly, if a lower-quality food source (warm-evolved algae) imposes selection for Daphnia genotypes that can better withstand this food environment, the stressful environment may erode the available genetic variation (Charmantier & Garant, 2005;Kause et al., 2001), resulting in a muted response of Daphnia to temperature selection (Box , Hypothesis 1 (a) versus (b)). Conversely, because zooplankton and phytoplankton tend to be exposed to similar abiotic environments in nature, it is also possible that we will see local adaptation, whereby cold-selected Daphnia evolve higher performance when they are fed with phytoplankton from a similar environment, and vice versa for warm-reared Daphnia (Box , Hypothesis 2, (c) versus (d)). Here, we have assayed thermal reaction norm evolution so that we can quantify the relative roles of phenotypic plasticity versus evolutionary change in Daphnia thermal responses.
We also examined whether the presence of predators mediated evolutionary responses of Daphnia to the thermal environment (Box ).
In the presence of predators, classical life-history theory predicts that consumers should evolve higher overall per capita growth rate (Abrams & Rowe, 1996) (Box , Hypothesis 3.1 (a) versus (b)). Secondly, Chaoborus larvae are known to be voracious predators of Daphnia, and thus, predator presence may reduce Daphnia population size (and thus genetic variation) to levels below which thermal evolution is feasible (Hypothesis 3.2 (a) versus (c)).  (Bold, 1949). After 4 weeks of acclimation to the laboratory environment, 30 Daphnia were haphazardly allocated to each of 160 glass jars (750 ml) filled with 700 ml COMBO media (Kilham, Kreeger, Lynn, Goulden, & Herrera, 1998). One thirdinstar Chaoborus was placed into half of the jars (n = 80). Half of the 160 jars were then placed in 12°C water baths (40 jars with predators, 40 without), and the other half were placed in 18°C water baths (40 jars with predators, 40 without). For the 18°C treatment, the temperature of the water bath was gradually raised from 12 to 18°C over a period of 1 week. To minimize environmental variation across water baths, each week all Daphnia population jars were haphazardly moved among water baths of the same temperature.

| Algae treatments and maintenance
Sixteen replicate populations of algae were seeded using the laboratory culture of Scenedesmus obliquus. Populations were grown in COMBO medium in 1.9-L glass jars. Jars were capped and aerated using filtered air (0.3-µm in-line HEPA-VENT filter) from a large aquarium air pump. Eight replicate jars were maintained at 12°C (cold-algae treatment), and eight were maintained at 18°C (warmalgae treatment).
Three times per week, approximately 3/5 of the contents of each of the eight algal populations per temperature treatment were poured out into a single bucket. Individual algal jars were refilled with COMBO and returned to the water bath. The pooled contents from each temperature treatment were fed to the appropriate Daphnia populations. The pooling of algae across the eight populations per temperature treatment for Daphnia feeding was purposeful because we wanted Daphnia to be exposed to generic "cold-reared" or "warm-reared" algae. Algal replicate jars remained separate for quantification of experimental evolution.
At four occasions during the experiment, we quantified algal cell size, and the number of algal cells after dilution, and immediately prior to the following dilution (typically 2 days later), using a FlowCam imaging particle counter (FlowCam ® VS Series; Fluid Imaging Technologies), at a flow rate of 0.3 ml/min. We calculated population growth rate during each period as cell number at time t + 1 divided by cell number at time t. There was no difference in population growth rate between algae maintained at the two temperature treatments (growth rate = 0.43 cells/day; temperature × date Box 1 Potential bottom-up effects of resource (algae) evolution on thermal reaction norm (TRN) evolution of consumers (Daphnia) F 1,124 = 0.17, p = 0.68). This lack of difference in growth rate is perhaps surprising because we typically believe that algae grow faster at warmer temperatures. We have since confirmed this result by growing the same strain of S. obliquus across multiple temperatures without dilution and found that differences in growth rate are not detectable until after the third day of continuous growth (data available upon request). Finally, at the beginning of the experiment we also used haemocytometer counts to confirm density estimates generated by the FlowCam, and found little difference between the estimates of cell density obtained using the two methods (FlowCam: 402,366 cells/ml; haemocytometer: 454,000 cells/ml).

| Daphnia maintenance
Daphnia were fed either cold-or warm-reared algae for the duration of the experiment. The number of algal cells per feeding was quantified using FlowCam, and we converted cells per feeding to carbon per feeding using published biovolume to biomass conversion equations (Montagnes, Berges, Harrison, & Taylor, 1994

| Common garden environment
On day 150 of the experiment, all predators were removed, and the temperature of all Daphnia and algae populations was set to 15°C (common garden environment Box 2 Potential top-down effects of predators (Chaoborus larvae) on thermal reaction norm (TRN) evolution of consumers (Daphnia)

| Assaying evolved differences in algal respiration and photosynthesis
After two rounds of algal replication in the common garden environment, we assayed cell volume and rates of photosynthesis and respiration, across a range of temperatures. We used the FlowCam to quantify average individual algal cell volume from a 0.3 ml sample from each replicate population. Scenedesmus obliquus exhibits morphological variation, whereby cells can either exist as unicells or colonies (coenobia, a group of daughter cells that share a parent cell wall) (Lürling & Van Donk, 1996). Our measure of cell volume does not distinguish between these two forms; we define "volume" as the total volume of material within the cell wall.
Rates of photosynthesis and respiration from subsamples of each replicate jar were assayed using a sealed glass microplate in combi- sensor spots were calibrated with air-saturated water (100% oxygen) and water containing 2% sodium sulphite (0% oxygen) at each experimental temperature. Algal cells were acclimated to the assay temperature for 1 hr in the dark prior to measurements. Three blank wells with COMBO medium were run at the same time as the phytoplankton at each temperature, and the average rate of oxygen flux in these wells was subtracted from the experimental wells to account for background microbial respiration. Gross photosynthesis was estimated as GP = net photosynthesis + respiration at each temperature. We assumed that net photosynthesis was directly proportional to oxygen production in the light. We mass-normalized rates of photosynthesis using the mean population biovolume (mean cell volume × cell density) from the source populations, measured using the FlowCam (flow rate = 0.3 ml/min) on the day of the respirometry assays. Cell volume (µm 3 ) was determined with area-by-diameter estimation (ABD) in each replicate population.

| Assaying evolved differences in Daphnia lifehistory traits
After 3 weeks in the common garden environment, four juvenile Daphnia from each replicate jar were placed in individual 50-ml Falcon brand centrifuge tubes. Once one of the four Daphnia matured and produced at least four neonates, the remaining three Daphnia were discarded. These four neonates were transferred to individual 50-ml

| Algal evolution
We used linear mixed-effects models (R package lmer) to evaluate whether S. obliquus photosynthesis or respiration rates evolved in response to rearing temperature. In these statistical models, selection temperature (12°C, 18°C) and assay temperature (10, 12, 16, 20, 24°C) were included as fixed effects, and "replicate" was included as a random effect. We assessed the statistical significance of the treatments and interactions using likelihood ratio tests, following Winter (2013). For the statistical analysis of algal cell volume, each replicate contributed one mean value to the analysis, and we used ANOVA to test whether rearing temperature explained variation in evolved cell volume.

| Daphnia population size and evolution
We used ANOVA to assess whether selection temperature (12°C, 18°C), predators (present/absent) and food type (cold vs. warm algae) explained significant variation in mean Daphnia population size.
We used linear mixed-effects models (R package lmer) to ex- Daphnia genotype (4 individuals) to the evolution assays, we included a "family" random effect in each of the models. We incorporated trait variation arising from drift among replicates by including random slopes and intercepts for the "family" term. This approach allows us to examine the variance in traits attributable to treatment effects after among-replicate variance is accounted for (Dutilleul et al., 2014).
We included all main effects and interaction terms in the full model for each dependent variable (main effects: food type, selection temperature, assay temperature, predator). We assessed the statistical significance of interaction terms using the ANOVA command in R (e.g., ANOVA model]), and we removed interactions terms if they were not statistically significant. We kept nonsignificant two-way interactions in the model if a three-way interaction that incorporated the two-way interaction was statistically significant. We had no a priori reason to expect significant statistical interactions between food type and predator presence, and none of these interactions were significant. All statistical analyses were conducted in R

| Cell size
When assayed after two generations in the common garden environment (15°C), algal populations that had been maintained for the duration of the experiment at the warmer temperature were 23% smaller in volume compared to algae maintained at the cooler temperature

| Daphnia population size
Both food type and predator treatment explained significant variation in Daphnia population size (Figure 2; predators: F 1,152 = 33.6, F I G U R E 1 Thermal reaction norms for Scenedesmus obliquus (a) photosynthesis and (b) respiration. There was no effect of selection temperature on photosynthesis. Algae reared at 18°C evolved higher rates of respiration than those reared at 12°C. Both photosynthesis and respiration increased with assay temperature. See Section 3.1.2 for statistics. Blue symbols denote algae selected at the cooler temperature, and red symbols represent algae selected at the warmer temperature. Error bars represent ± 1 standard error

| Body size
Algal food type altered Daphnia body size evolution in response to selection temperature ( Figure 3a and The presence of predators modified how Daphnia body size evolved in response to the thermal environment ( Figure 3b and Table 1). The effect of temperature selection was more pronounced when Daphnia were also reared with predators, with Daphnia evolving overall larger bodies in the 12°C versus 18°C selection treatments.

| Time to maturity
Variation in Daphnia development time was primarily explained by assay temperature, with Daphnia assayed at warmer temperatures maturing several days earlier than those assayed at cooler temperatures ( Figure 4a and Table 1). There was no effect of selection temperature or algal food type on TRN evolution. There was also no effect of predator presence on Daphnia development time TRNs (Figure 4b).

| Per capita growth rate, r
Algal food type altered how TRNs for Daphnia r evolved in response to selection temperature ( Figure 5a and Table 1). When fed cold-evolved algae, Daphnia evolved steeper TRNs in the 12°C selection treatment, but when fed warm-evolved algae, Daphnia evolved steeper TRNs in the 18°C selection treatment (Figure 5a F I G U R E 2 Effect of food type, predators and selection temperature on mean Daphnia population size. Daphnia fed coldevolved algae maintained a 12% higher overall population size compared to those fed warm-evolved algae, and predators reduced Daphnia population size by 22% (see Section 42.3.3 for statistics).
Error bars are ± 1 standard error F I G U R E 3 Effect of (a) algal food type and assay temperature, and (b) predators and assay temperature on Daphnia body size thermal reaction norm evolution. (a) Daphnia fed cold-evolved algae diverged evolutionarily in response to selection temperature. However, there was no effect of selection temperature when Daphnia were fed warm-evolved algae. (b) Daphnia reared in the presence of predators evolved larger bodies in the cold selection environment. In the absence of predators, there was no difference in body size TRN between the two temperature selection treatments. See Table 1 for statistics, error bars are ± 1 standard error and Table 1). Assay temperature explained a significant fraction of variation in Daphnia r, with Daphnia exhibiting faster r when grown in warmer assay temperatures. There was no effect of predators on TRN evolution in Daphnia r (Figure 5b and Table 1).

| D ISCUSS I ON
The We discuss these results in further detail below.

| Algal evolution
Algae exhibit cell size plasticity in response to temperature (Chalifour & Juneau, 2011;Chen et al., 2011;Margalef, 1954;Rhee & Gotham, 1981), and thus, differences in cell size between the two temperature selection experiments were likely a function of both plasticity and evolution. The data collected here do now allow us to quantify exactly when in the experiment algal evolution occurred. However, using population sizes less than those used in this experiment Two recent studies on thermal reaction norm evolution in other species of phytoplankton also reported the evolution of elevated respiration in warmer selection environments, but in both of these studies, they also saw the evolution of increased photosynthesis (Chlorella vulgaris: Padfield et al., 2016, Chlamydomonas reinhardtii: Schaum et al., 2017. The duration of selection was 100 generations, and 10 years, for Chlorella, and Chlamydomonas, respectively, suggesting that the 70 generations used here may have been too short to observe photosynthesis evolution. However, the evolved increase in resource consumption (respiration) but not resource acquisition (photosynthesis) at warmer temperatures is consistent with the observed evolved decrease in cell size. Finally, although not directly tested here, cell size is positively correlated with lipid content in other algal species (Bono, Garcia, Sri-Jayantha, Ahner, & Kirby, 2015). This pattern suggests a possible mechanistic link between cold temperature, cell size, increased algal fatty acids and increased nutritional quality of cold-reared algae.

| Effect of resource thermal evolution on Daphnia population size
Our results demonstrated that algal thermal evolution had significant effects on Daphnia population size. Daphnia maintained on cold-reared algae attained a 12% higher population size during the experiment compared to Daphnia maintained on warm-reared algae.
These results are consistent with a previously published study showing that cold-reared Scenedesmus obliquus are more nutritious and sustain higher Daphnia population sizes, compared to warm-reared S. obliquus (Breuer et al., 2013;Sikora et al., 2014;Vigeolas et al., 2012).

| Effect of resource evolution on Daphnia thermal evolution: "bottom-up effects"
From previous studies demonstrating a positive correlation between environmental quality and genetic variation/heritability (Bochdanovits et al., 2003;Charmantier & Garant, 2005), we predicted that Daphnia reared long term on cold-evolved (higher quality) phytoplankton would show an elevated evolutionary response to temperature-mediated selection (Box , Hypothesis 1). However, we also predicted that because zooplankton and phytoplankton often encounter the same abiotic environment in nature, a "better" environment might be one in which the consumer environment matches that of the resource (Box , Hypothesis 2). Our data were consistent with both hypotheses. (a) Daphnia body size evolution in response to selection temperature was more pronounced in populations that were fed the cold-evolved/higher quality algae, consistent with Hypothesis 1 (Figure 3a); (b) conversely, Daphnia r evolution exhibited patterns indicative of local adaptation between the consumer and resource thermal environment, consistent with Hypothesis 2 (Figure 5a). What we did not predict at the outset of this study was that rapid algal evolution could potentially change the slope of

| Potential for maladaptation
One would predict that organisms with rapid generation times (e.g., phytoplankton or microbes) will evolve much more quickly F I G U R E 5 Effect of (a) algal food type and assay temperature, and (b) predators and assay temperature on Daphnia r thermal reaction norm evolution. At the warmer assay temperatures, Daphnia reared at 12°C evolved higher r when fed cold-evolved algae, and Daphnia reared at 18°C evolved higher r when fed warm-adapted algae ( to warming, relative to their consumers (e.g., zooplankton), which have longer generation times and smaller population sizes. As temperatures in nature continue to warm, the mismatch in rates of thermal evolution between resources and consumers could result in slower-evolving consumers interacting with resources that have already evolved to a changed environment. This hypothetical situation is not totally dissimilar from the "warm-evolved algae + coldselected Daphnia" treatment combination used in this study. For both body size and r, Daphnia in this treatment combination evolved muted thermal plasticity compared to Daphnia in the "cold-evolved algae + cold-selected Daphnia" combination. If thermal plasticity confers fitness benefits, then rapid thermal adaptation in resources could lead to maladaptation in consumers.

| Effect of predators on Daphnia thermal evolution: "top-down" effects
We predicted that the presence of predators would modify Daphnia TRN evolution (Box , Hypothesis 3). We did not observe the predicted increase in Daphnia r in response to predators, but Daphnia did evolve larger bodies, especially those in the 12°C selection treatment. These results are consistent with a previous study showing the evolution of larger bodies and earlier reproduction in the same predator-prey system (Spitze, 1991). However, another study that used the same study system, but with food limitation, did not observe the evolution of larger Daphnia body size in the presence of Chaoborus predators (Tseng & O'Connor, 2015).

| CON CLUS IONS
Using laboratory experiments on a tri-trophic aquatic community, we have demonstrated that species interactions can play an important role in determining how a focal population response to temperature-mediated selection. The challenge moving forward is to apply these results to natural settings and to assess how frequently species interactions hinder or help focal populations adapt to warming climates. There are multiple cases in nature where interacting species may respond to the same selection pressure but at very different speeds (e.g., insects and their coevolved viruses (Cory & Myers, 2003), small mammals and their prey (hare/lynx: Krebs, Boonstra, Boutin, & Sinclair, 2001), trees and their seed predators (Mezquida & Benkman, 2014), fish and zooplankton (Vanni, 1986)), and thus, we expect the "bottom-up" results of this study to be widely applicable across natural food webs.

ACK N OWLED G EM ENTS
We thank K. Hofer, S. Cao and M. Urquhart-Cronish for help with data collection and K. Beall for maintaining the temperamental environmental chambers. We thank M. O'Connor for the use of the O'Connor laboratory respirometer and FlowCam.

CO N FLI C T O F I NTE R E S T
None declared.

DATA ACCE SS I B I LIT Y
All raw data are available in Dryad (https ://doi.org/10.5061/dryad. d6f6q7p) .