Costs and trade-offs of grazer-induced defenses in Scenedesmus under deficient resource

The green alga Scenedesmus obliquus can form inducible defensive morphs under grazing threat. Costs and trade-offs of inducible defense are expected to accompany the benefits of defensive morphs, but are hard to detect under nutrient-sufficient experimental conditions. To test the existence of costs associated with inducible defense, we cultured S. obliquus along resource availability gradients in the presence or absence of infochemical cues from Daphnia, and measured the strength of defensive colony formation and fitness characters. Under the lowest phosphorous concentration, the expression of inducible defensive colony resulted in decreased growth rate, which provides direct evidence for physiological costs. Along the gradient reduction of phosphorous concentration or light intensity, inducible defense in S. obliquus showed a decreasing trend. However, the photosynthetic efficiency of S. obliquus was barely affected by its defense responses, suggesting that the negative correlations between resource availability and colony formation of this alga may be due to resource-based trade-offs in the allocation of limited resources. Thus, our results indicated that expression of inducible defense of S. obliquus was impaired under insufficient phosphorus or light. Furthermore, under severe phosphate deficiency, obvious physiological costs of inducible defense could be detected even though defensive colony formation also decreased significantly.

Scientific RepoRts | 6:22594 | DOI: 10.1038/srep22594 In terms of energy requirements, defenses that are induced only in the presence of herbivores are considered more efficient than constitutive defenses under variable grazing risk 14 . The adaptive benefits of colony formation for Scenedesmus spp. have been investigated in many studies 15,16 . However, no direct evidence about the cost accompanying with the benefit is detected. Trade-offs and costs associated with grazer-induced colony formation exist theoretically; otherwise, this defensive phenotype would be fixed 17 . Since the energy storage of organisms is limited, excessive investment of energy for defense against grazers would impose restriction on growth and maintenance. For Scenedesmus, two main potential indirect costs coupled with morphological defense are identified: reduction of surface-to-volume ratios and enhancement of sinking risk 18 . Both costs would result in loss of competitive advantage in phytoplankton populations. As four, eight, or even more algal cells join together to form a larger resistant coenobium, volume is increased compared to single algal cells, but a high surface-to-volume ratio is sacrificed. In addition, overlapping or staggered coenobiums may result in "package effect" on inner cells, and light capture and nutrient assimilation may be negatively influenced 19 .
Apart from the relatively small surface area for nutrient and light absorption, increased sedimentation rate of large defensive coenobiums also have negative effect on the population of Scenedesmus. Colonies not only settle faster but are also less able to resuspend to upper water compared to unicells. For non-motile phytoplankters, prolonged residence in warm upper water with sufficient illumination is a primary requirement. Higher risk of sinking out of euphotic zone increases the loss of population biomass, and decreases the growth rate for light limitation 18,20 . Aside from grazing, sedimentation loss appears to be a major threat to small phytoplankton in the pelagic zone 21 . Thus, to preserve high biomass, a trade-off in production of stable large colony may be required to avoid the selective pressure of predation and division into single cells to maximize nutrient uptake in Scenedesmus.
Recent research has tried to reveal the cost associated with induced colony formation in Scenedesmus 18,22,23 . However, the direct fitness costs of forming defensive colonies in terms of growth or other physiological indexes 18,24 has been difficult to determine, which may be due to the abundance of nutrients in culture medium used in those studies. Because of the package effect 19 , colonies should have less chlorophyll-specific absorption coefficients than unicells, which should be reflected in the growth rate. However, no significant decrease is observed in growth rate 18 . Forming defensive morphs tends to improve the adaptive fitness of Scenedesmus. Nevertheless, when alga is cultured in poor conditions, such as nutritional deficiency or light insufficiency, such conditions may be disadvantageous if the cost of maintaining colonies is high. Colony formation of S. acutus has been reported to be a constitutive defense response under phosphorus-sufficient conditions, but an inducible defense response under P-limited conditions 22 , indicating the existence of trade-off between colony formation and nutrient availability. As previous studies have showed, maintaining defensive morphs impairs the competitive ability of Scenedesmus under nutrient-limited conditions caused by competition 23,25 . Therefore, we hypothesized that when resource is insufficient, the cost of induced defense may be obvious, and a trade-off between defense and resource absorption may emerge. To examine our hypothesis, we measured the induced colony formation, population growth, and photosynthesis of Scenedesmus under different phosphorus concentrations and light intensities.

Inducible defense of Scenedesmus under different phosphorus concentrations. Addition of
Daphnia filtrate stimulated the increase of colony size under all five P concentrations (Fig. 1). In the absence of Daphnia filtrate, no difference was detected in cells per particle among different P concentrations (F = 1.172, p = 0.331). A three-way ANOVA indicated that the induced colony formation was significantly decreased by decrease in P concentration (Table 1). Under the exposure of Daphnia infochemicals, the colony size of S. obliquus increased dramatically and peaked on day 4 ( Fig. 1). Fitted curves revealed that the value of C max declined, which was caused by decreased P concentration. To investigate the relationship of colony size with resource availability, the observed C max from each triplicate of the Daphnia filtrate treatments was fitted against P concentrations or light intensities (Fig. 2). The dots of C max -observed were the actual maximums. The C max -observed in response to P concentration followed a rectangular hyperbolic model and increased with increasing P concentration, but it no longer increased when the P concentration was above 0.5 mg L −1 (Fig. 2a).
The growth capacity of S. obliquus declined with decreased P concentration (Fig. 3). Addition of Daphnia filtrate led to significant decline in growth rate in the lowest P concentration (F = 9.188, p = 0.039). In other P concentrations, no statistically significant decrease was observed between control and Daphnia filtrate groups. A two-way ANOVA was conducted to test the effect of Daphnia filtrate and P concentration on growth rate. The result indicated that both factors had significant influences on the growth rate, but no interaction was detected between the two factors ( Table 2).
The photosynthetic efficiency of S. obliquus was affected by different P concentrations with or without Daphnia filtrate. The maximal efficiency of PSII photochemistry (F v /F m ) of S. obliquus was not significantly different between the control and Daphnia filtrate groups (F = 0.0117, p = 0.914). A three-way ANOVA revealed significant positive effects of time (F = 5.475, p = 0.002) and P concentration on F v /F m (F = 7.115, p < 0.001). Consistent with the ratio of F v /F m , Daphnia filtrate had no significant influence (F = 0.756, p = 0.387) on the effective quantum yield of PSII (Φ PSII), but Φ PSII varied significantly with both time (F = 7.283, p < 0.001) and P concentration (F = 6.648, p < 0.001). In the two low P concentration treatments (0.05 and 0.2 mg L −1 ), Φ PSII decreased from day 5. ETR max showed an increase on day 3. No significant variation was observed between control and Daphnia filtrate groups.  Table 1). The number of cells per particle was fitted by Gaussian distribution under five light intensities. An increasing trend of the inducible defensive colony formation was observed with increased illumination (Fig. 4). Unlike the hyperbolic relationship between C max -observed and P concentration, C max -observed in response to light intensity matched the linear models (Fig. 2b). One-way ANOVA on C max indicated significant light intensity effect (F = 22.928, p < 0.001). C max of S. obliquus at 9 μmol photons m −2 s −1 was less than half of C max at 72 μmol photons m −2 s −1 . However, no significant difference of C max was observed under the light intensities within the range of 18-54 μmol photons m −2 s −1 . The decreased growth rate of S. obliquus at 9 μmol photons m −2 s −1 indicated growth inhibition by low light intensity (Fig. 5). Results of two-way ANOVA showed that different light intensities significantly affected the growth rate of S. obliquus, but the addition of Daphnia filtrate had no significant influence ( Table 2). The growth rate of S. obliquus under the highest P concentration in the different P gradient experiment (at the light level of 45 μmol photons m −2 s −1 ) was lower than that in the light intensity experiment. This was possibly due to the higher initial cell density in the P gradient experiment, and the decrease in intracellular phosphorus quota because of pre-treatment of algal cells in non-phosphorus medium.
The photosynthetic efficiency of S. obliquus changed with light intensity with or without Daphnia filtrate. Three-way ANOVA indicated that the ratio of F v /F m was influenced significantly by time (F = 4.261, p = 0.008) and light intensity (F = 5.017, p = 0.001). The ratio of F v /F m increased with increase in light intensity. At the highest light intensity, the initial F v /F m was high, but decreased with cultural time. Moreover, a time × light intensity interaction (F = 1.946, p = 0.041) was found. No significant Daphnia filtrate effect was detected (F = 0.702,    Vertical lines represent ± 1 SE (n = 3). Different capital letters above solid bars denote significant (P < 0.05) differences among the control groups, while different lowercase letters above hollow bars denote significant (P < 0.05) differences among Daphnia filtrate groups. The asterisk above a short horizontal line indicates significant difference (P < 0.05) between control and Daphnia filtrate groups under certain P concentrations.

Discussion
In this study, we investigated the grazer-induced morphological defense of S. obliquus and the effect of resource gradient on this defense response. At all phosphorus concentrations and light intensity treatments, Daphnia filtrate induced the defensive colony formation of by S. obliquus. Consistent with previous studies 18,26 , the numbers of cells per particle increased under the exposure of Daphnia filtrate and then decreased, possibly because of the increase of cell densities and the degradation of chemical cues. Colony formation in response to Daphnia infochemicals exhibited a decreasing trend with decreasing environmental resource availability (Fig. 2). A strong two-way interaction between resource availability (phosphorous concentration or light intensity) and Daphnia filtrate indicated that the character of phenotypic plasticity of S. obliquus was determined by the complex interactive effects of biotic and abiotic factors, which is consistent with previous studies. Alteration of anti-predation response depending on environmental resource availability has also been observed in many other species of photoautotrophs [27][28][29] and preys at higher trophic levels 30,31 . The formation of defensive colony is highly effective for withstanding grazing pressure and preserving biomass 10,32 . However, limitations and costs should be expected, otherwise colony formation will be a constitutive defense 33 . To avoid excess costs of defense, many organisms adjust the intensity of inducible defenses based on predation risk and conspecific density 30,34,35 , indicating trade-offs between the benefits and costs of inducible defense. The "plant defense hypotheses" 36 posits that decline of defensive colony formation along the resource availability gradient. In the present study, the decrease in growth rate under low resource availability condition indicates that inducible defense is accompanied by some limitations and costs.
Both light limitation and phosphate deficiency impaired the grazing-induced defense in S. obliquus. The negative correlations between resource availability and colony formation in S. obliquus may be due to resource-based trade-offs in the allocation of limiting resources. It has been already known that extracellular polysaccharides play an important role in increasing the stickiness of algal cells to form coenobia and to aggregate together. Biosynthesis of polysaccharides requires more energy and carbohydrate allocation. Therefore, increasing the production of extracellular polysaccharides will require extra investment from the intracellular pools of carbon. Limitation of external P i (inorganic phosphate) will reduce the intracellular concentrations of P i and ATP synthesis 37 . For PSII, the electron flow and carbon fixation rate in the algae exhibit linear correlation 38 , and the decrease of PSII electron transport rate will inhibit the biosynthesis of polysaccharides. Deficiency of external inorganic phosphate supply affects the photosynthetic apparatus of phytoplankton 39 because a part of the light energy is allocated for nutrient uptake instead of carbon fixation 40 . With P i deficiency, photosynthesis is inhibited because of decreased ribulose-1,5-bisphosphate pool size 41 , and chlorophyll content is also lower 42 , thereby causing decreased carbon assimilation area in algal cells. In the present study, P-limitation had a significant negative effect on the ratio of F v /F m and value of Φ PSII, probably suggesting stress on PSII reaction centers under low P concentrations. ETR max was also decreased under low P concentrations, suggesting that efficiency of the electron transport chain was affected. Although no significant Daphnia filtrate effect was observed on photosynthetic parameters, growth reduction emerged under phosphorous deficiency, specifically under the lowest P concentration, which is a relatively low level in eutrophic lakes 43 . Growth inhibition of induced algae was possibly due to increased requirement of extracellular polysaccharides to form defensive colonies under phosphate deficiency, suggesting the direct costs associated with inducible defense of S. obliquus.
The influence of light intensity on the defensive colony formation of S. obliquus was similar to that of phosphorous concentration gradient. In this study, the inducible colony formation decreased significantly at a light intensity of 9 μmol photons m −2 s −1 (Fig. 4), indicating that the ability of S. obliquus to form defensive colonies Vertical lines represent ± 1 SE (n = 3). Different capital letters above solid bars denote significant (P < 0.05) differences among control groups, while different lowercase letters above hollow bars denote significant (P < 0.05) differences among Daphnia filtrate groups.
was inhibited under light limitation, which may be due to decreased production of carbohydrate when light intensity is low 44 . Biomass growth, carbon fixation rate, and carbohydrate productivity of phytoplankton increase with increasing light intensity below the light saturation point 45 . Under low light intensity, algal cells accumulate more lipid than carbohydrates, whereas high light intensity causes carbohydrate accumulation and cellular lipid content decreases 46 . As the key factor of photosynthesis, light has a major influence on photosynthetic carbon fixation, energy synthesis, and nutrient consumption rate 47 . The observed decrease in Φ PSII affected by Daphnia filtrate indicated the potential package effect of forming colonies. ETR max was notably low under the lowest light intensity, which indicated strong inhibition of electron transport efficiency. Biomass production was reduced due to decrease of photosynthetic efficiency, and no efficient energy or carbohydrate source was available to maintain biomass growth for defensive colony formation. Therefore according to our study, despite the high selection pressure of herbivores, deficiency of resource availability is likely to be the limitation for S. obliquus in forming defensive colonies, and trade-offs between defense and development will occur. Furthermore, the reduction of photosynthetic efficiency of S. obliquus under low resource conditions may limit energy allocation for defense.
However, contradictory evidence on the trade-offs of resource allocation between growth, reproduction, and defense has been reported 28,[48][49][50][51][52][53] . The resource availability hypothesis predicts that large investment on anti-herbivore defenses is favorable when resources are limited 48,49 . However, the carbon-nutrient balance hypothesis assumes resource allocation between production of defensive compounds and plant growth 50 . The variations of allocation costs associated with inducible defense may depend on the kind of deficient resources 28,51 , type of defense (different defensive mechanisms may have different potential sources of costs 52 ), and biosynthesis of defensive compounds 53 . Considering the defensive mechanism of S. obliquus, poor availability of phosphorous and light will have negative effect on colony production. Colony formation in S. obliquus induced by FFD-6, a type of anionic surfactant, also decreases in nutrient-limited cultures 54 . Unlike S. obliquus, S. acutus produces colonies constitutively when phosphorous is sufficient despite Daphnia cues, but only produces colonies after perceiving Daphnia cues when phosphorus is limited 22 , indicating that the cost of forming colonies may be high and that nutrient availability does affect colony formation. Although allocation of limited resources (phosphorus and light) to growth and reproduction instead of resistance would be favored by S. obliquus in order to avoid physiological cost, maintenance of defensive colony under phosphorus deficiency decreased S. obliquus growth. Growth reduction was also detected in Phaeocystis globosa which formed grazer-induced colonies in low nutrient conditions, indicating the existence of costs 29 .
Despite the internal physiological costs of forming anti-grazing morphs of S. obliquus, previous studies also indicated the existence of other costs 23,25 . Maintenance of colonies reduces the ability to compete for resources and habitats with competitors 23 . Our previous study also showed that the co-existence of competitors impairs the inducible defense in S. obliquus 25 . Thus, these experimental studies have demonstrated the opportunity costs of defense. However, under natural conditions with complex abiotic and biotic interactions, potential ecological costs may also occur. Ecological costs are indirect consequences of induced defense and are difficult to test under laboratory conditions. Research has begun to detect the ecological costs of plants and animals (e.g., delayed flowering 55 , attracting additional natural enemies 56 , trade-offs between resistance and tolerance 57 , and other kinds of ecological costs). For phytoplankton, forming large-sized colonies is an efficient morphological defense against small herbivores. However, this defensive morph will be less effective when encountering grazers with large gape size 15 . The enhanced sinking rate of colonial S. obliquus will be disadvantageous for the alga to remain suspended in upper warm and euphotic waters 10 .
In conclusion, the expression of inducible defensive colony in S. obliquus shows dynamic responses on the availability of main environmental resources (phosphate and light). Resource insufficiency limits the expression of inducible defensive colony formation, thereby suggesting the existence of trade-offs of intercellular resource allocation between defense and growth. Our observation of growth inhibition of defensive algae under low phosphorus condition suggests that the costs may be detected under specific conditions, such as nutrient deficiency. Studying the costs accompanying the benefits of inducible defense is highly important for understanding the mechanism and evolution of inducible defense. Our work provides evidence supporting the existence of costs and trade-offs of inducible defense in S. obliquus.

Methods and Materials
Algal culture. The freshwater green alga Scenedesmus obliquus FACHB-416 (Chlorococcales, Chlorophyceae) can distinctly respond to infochemicals from grazers, such as cladoreans Daphnia spp. and rotifers. This alga is a common species used for inducible defense studies. S. obliquus was cultured with a modified algal growth medium BG-11 in a climate-controlled chamber at a constant temperature of 25 ± 0.5 °C with a light-dark period of 14:10 h.
Preparation of grazer infochemicals. The chemical cues that can induce anti-grazing colony formation in Scenedesmus exist in the water of herbivorous zooplanktons such as Daphnia, which feeds on Scenedesmus 5,6 . Therefore, addition of filtered water from Daphnia culture was used as a common method to induce colony formation. In our experiments, the grazers Daphnia magna from a laboratory clone was incubated at a density of 300 individuals per liter for 24 h feeding on S. obliquus at a density of 10 5 cells ml −1 . Then, the water was filtered through a 0.1 μm membrane filter (Millipore Corporation, USA) to eliminate bacteria and other impurities. Another suspension of S. obliquus without D. magna was filtered as control water. In preparing both control and D. magna water filtrates, phosphate was first excluded during culture medium preparation, and was added back at different concentrations after feeding for 24 h as outlined below.
Experimental procedures. Five phosphorus concentrations and five light intensity levels were established to investigate the potential costs and trade-offs of forming defensive colonies. In the experiment involving Scientific RepoRts | 6:22594 | DOI: 10.1038/srep22594 different phosphorus (P) concentrations, modified BG-11 culture media with ammonium as nitrogen source were prepared at 0.05, 0.2, 0.5, 1.0, and 5.0 mg L −1 of P. K 2 HPO 4 was applied as phosphorous resource and KCl was added to compensate for potassium shortage. The measured initial P concentrations in control groups were 0.049 (± 0.009), 0.198 (± 0.032), 0.515 (± 0.029), 1 The total phosphorus and nitrogen concentrations were detected by the ammonium molybdate spectrophotometric method and the alkaline potassium persulfate digestion UV spectrophotometric method (China SEPA) 58 . The experimental cultures were maintained at 25 °C and illuminated at normal light level of 45μmol photons m −2 s −1 using fluorescent lights on a 14:10 light:dark cycle. Before the experiment, algal cells were starved in non-phosphate conditions for 3 days to eliminate the interference of intracellular phosphate quota. Starved algal cells were then washed with non-phosphate BG-11 medium, centrifuged at a speed of 662 g for 20 min, and placed into 250 ml Erlenmeyer flasks with different phosphorous concentrations. The initial cell density was about 7.0 × 10 4 cells ml −1 .
In the experiment involving different light intensities, culture flasks were placed separately in climate-controlled chambers at five levels of light intensity, i.e., 9,18,36,54, and 72 μmol photons m −2 s −1 . Algal cultures were accommodated at respective light intensities for 3 days in advance 59  Experimental measurements. Samples (1 ml) of every control and Daphnia filtrate-treated groups were taken at 9:00 a.m. every day under aseptic conditions. Lugol's iodine solution was added at 2% as preservative after sampling. Two major quantitative traits, namely, population growth and colony size, were measured. Population growth was calculated based on cell density, which was counted using a hemocytometer (0.1 mm deep) on an Olympus microscope (Olympus 6V20WHAL; Tokyo, Japan). Eighteen 0.1 mm 3 count areas of one sample were observed and analyzed. Growth rate was obtained as the mean slope of ln (S. obliquus cells ml −1 ) over time. The mean numbers of cells per particle were also calculated from the above counts (cell abundances and cells in different morphological particles). The number of cells per particle (C) versus cultural time (t) at all P concentrations and light intensities were fitted by using a Gaussian distribution: where C max is the maximum number of cells per particle, t 0 is the time to reach C max , and a is the full width at half maximum intensity of the curve. To determine the relationship between resource availability and defensive colony formation, the observed number of C max was plotted against P concentration and light intensity. Samples (2 ml) of each group were taken every day for the measurement of photosynthetic efficiency with the PHYTO-PAM phytoplankton analyzer (Heinz Walz GmbH, Effeltrich, Germany). The maximal efficiency of PSII photochemistry was determined as F v /F m , where F v = (F m − F 0 ), and F m and F 0 are the maximal and minimal chlorophyll fluorescence yield of a dark-adapted suspension, respectively. The effective quantum yield of PSII (Φ PSII) was calculated according to the following expression: (F' m − F s )/F' m , where F s and F' m are the stable and maximal chlorophyll fluorescence in light-acclimated algal suspensions, respectively. The electron transport rate (ETR) versus irradiance (PAR) curve was plotted for 20 different PARs within the range of 0-2000 μmol protons m −2 s −1 . The maximal electron transport rate (ETR max ) was determined from a curve after fitting it to the model proposed by Platt et al. 60 .

Statistical analyses.
All values are presented as mean ± SE. Growth rates and numbers of cells per particle were compared by two-way ANOVA with P concentration or light intensity and Daphnia filtrate as the fixed factors. Three-way ANOVA was used to compare the differences and interactions among different environmental resource levels (P or light separately), Daphnia filtrates, and time. All data were analyzed using SigmaPlot 11.0.