Growth Rate Hypothesis does not apply across colimiting conditions: cholesterol limitation affects phosphorus homoeostasis of an aquatic herbivore


Corresponding author. E-mail:


1. Herbivores show stronger control of element homoeostasis than primary producers, which can lead to constraints in carbon and nutrient transfer efficiencies from plants to animals. Insufficient dietary phosphorus (P) availability can cause reduced body P contents along with lower growth rates of animals, leading to a positive relationship between growth and body P.

2. We examined how a second limiting food component in combination with dietary P limitation influences growth and P homoeostasis of a herbivore and how this colimitation influences the hypothesized positive correlation between body P content and growth rates. Therefore, we investigated the responses in somatic growth and P stoichiometry of Daphnia magna raised on a range of diets with different amounts of P and the sterol cholesterol.

3. Somatic growth rates of D. magna increased asymptotically with increasing P as well as with increasing cholesterol availability. The body P content increased with increasing dietary P and stabilized at high dietary P availability. The observed plasticity in D. magna’s P stoichiometry became stronger with increasing cholesterol availability, i.e. with decreasing colimitation by cholesterol.

4. At P-limiting conditions, the positive correlation between body P content and growth rate, as predicted by the growth rate hypothesis (GRH) applied to the within-species level, declined with increasing cholesterol limitation and disappeared entirely when cholesterol was not supplied. Thus, even when Daphnia shows no growth response owing to strong limitation by the colimiting nutrient, the body P content may vary substantially, calling into question the unconditional use of herbivores’ P content as predictor of a potential P limitation in nature.

5. The observed interaction between dietary P and cholesterol on Daphnia’s growth and stoichiometry can be used as a conceptual framework of how colimiting essential nutrients affect herbivore homoeostasis, and provide further insights into the applicability of the GRH within a consumer species.


One major issue in nutritional ecology is the dietary mismatch on the plant–animal interface (Elser et al. 2000a; Hessen et al. 2004). In contrast to herbivores, primary producers often exhibit large variations in their elemental and biochemical concentrations, leading to imbalanced diets and ensuing negative effects on herbivore performance (e.g. Müller-Navarra 1995; Sterner et al. 1998; Raubenheimer & Simpson 2004). Thus, besides the amount of available food, food quality constraints on the plant–animal interface can be a key to understanding restricted transfers of carbon and nutrients to organisms of higher trophic levels (Sterner & Hessen 1994; Elser et al. 2000a). Several studies have shown that herbivore growth and performance can be negatively affected by imbalanced ratios of macronutrients, e.g. proteins, lipids and carbohydrates (e.g. Raubenheimer & Simpson 2004), or by imbalanced element (N, P or Ca) to carbon (C) ratios (Sterner & Elser 2002). Furthermore, essential micronutrients such as polyunsaturated fatty acids (PUFA), amino acids (AA), vitamins or sterols might be limiting food components for herbivores (Von Elert, Martin-Creuzburg & Le Coz 2003; Anderson, Boersma & Raubenheimer 2004; Müller-Navarra 2008).

The limitation of herbivore production caused by element deficiency is a pivotal issue in ecological stoichiometry that deals with the balance of energy and multiple chemical elements in ecological interactions (Elser et al. 2000b). Ecological stoichiometry is a useful tool to analyse how the element composition of organisms affects production, nutrient cycling and food web dynamics (Sterner & Elser 2002). In freshwater systems, especially in lakes, phosphorus (P) is often the main limiting element, which can lead to reduced herbivore production because of low food quality, i.e. high C/P ratios of the food (Sterner & Elser 2002). Stoichiometric theory assumes element homoeostasis of consumers and predicts a growth limitation when the ratio of the element in the food drops below a certain threshold (Sterner & Hessen 1994; Hessen et al. 2004). The threshold elemental ratio for P (TERC/P) depends on species-specific body P contents and gross growth efficiencies, indicating different P requirements between species (Sterner & Hessen 1994; Frost et al. 2006). The growth rate hypothesis (GRH) provides a framework to explain differences in P requirements among species and states that the P content of consumers is directly related to their ribosomal RNA concentration (Elser et al. 2000b). As an enhanced allocation to P-rich ribosomal RNA is required to increase protein synthesis and thus growth, the GRH predicts that species with high body P contents show higher maximum growth rates (Elser et al. 2003). The positive relationship between growth, body P content and RNA, originally proposed for comparison across species, may apply only to P-saturating conditions where species acquire sufficient P for the synthesis of ribosomal RNA. At P-limiting conditions, the allocation of P to ribosomal RNA synthesis might be restricted, which can cause reduced body P content as well as reduced growth of an organism. Thus, the GRH might hold within the species level when an organism shows P-limited growth along with plasticity in body P stoichiometry. Repeatedly, it was shown that herbivores growing on P-limiting diets had lower body P contents and are therefore not as strictly homoeostatic as previously believed (Frost & Elser 2002; DeMott & Pape 2005; Ferrão-Filho, Tessier & DeMott 2007). Additionally, several studies reported positive relationships between growth rate and body P content under P limitation as predicted by the GRH within single consumer species (DeMott, Gulati & Siewertsen 1998; Schade et al. 2003; Fink & Von Elert 2006), and other studies showed a decoupling of this relationship when P was not limiting for growth (Elser et al. 2003; Acharya, Kyle & Elser 2004; Shimizu & Urabe 2008). The latter studies showed an influence of other nutritional factors on the relationship between growth rate and body P content. Thus, the actual question is how strongly the growth rate of an organism and its degree of homoeostasis depend on factors other than the nutritional availability of the limiting element.

Besides important factors, such as food abundance, ingestibility and digestibility of the food, potentially limiting micronutrients (e.g. PUFA, AA and sterols) may interact with element limitation and thereby influence organism’s stoichiometry. A colimitation of growth by another nutrient might release the organism from physiological P limitation even under potentially P-limiting conditions. The organism requires less P because of the restricted growth caused by the colimiting nutrient, which facilitates the ability to maintain homoeostasis, i.e. attenuates the decline in body P content under P-limiting conditions. Recent laboratory studies showed the colimitation of herbivore growth by P and PUFA (Becker & Boersma 2003) or sterols and PUFA (Martin-Creuzburg, Sperfeld & Wacker 2009; Martin-Creuzburg, Wacker & Basen 2010), but data on organism’s stoichiometry are lacking. Sterols are essential food components for herbivorous arthropods (Behmer & Nes 2003; Von Elert, Martin-Creuzburg & Le Coz 2003), which have to metabolize dietary phytosterols into cholesterol (Svoboda & Thompson 1985). Cholesterol, as an indispensable component of plasma membranes, plays a role in membrane-specific temperature adaptation (Crockett 1998) and serves as a precursor for moult-inducing ecdysteroids in arthropods (Goad 1981). Edible cyanobacteria can cause sterol-limited growth of herbivorous crustaceans (Von Elert, Martin-Creuzburg & Le Coz 2003) because they usually lack sterols (Volkman 2003), enhancing the possibility of sterol limitation during cyanobacterial blooms. Although cyanobacteria are often dominant in eutrophic lakes with high total P concentrations, high C/P ratios during cyanobacterial blooms are possible because of a strong carbon fixation of the autotrophs relative to their P acquisition (DeMott & Gulati 1999; Hessen, Van Donk & Gulati 2005), which can lead to P-limited growth of herbivorous crustaceans (DeMott, Gulati & Van Donk 2001). Cyanobacterial blooms are more likely to occur at elevated water temperatures during summer (Paerl & Huisman 2008), conditions where crustaceans need more cholesterol for temperature adaptation (Hassett & Crockett 2009; Sperfeld & Wacker 2009). Thus, the higher cholesterol demand of herbivorous crustaceans coincidently with the (phyto)sterol shortage caused by cyanobacteria dominance might lead to a colimitation of herbivore growth by dietary P and sterols. Also, a dominance by other phytoplankton species might lead to a sterol limitation of herbivorous crustaceans, because high light intensity and low P availability can decrease the sterol concentration of eukaryotic algae (Piepho, Martin-Creuzburg & Wacker 2010).

In laboratory experiments, we tested how the colimitation by dietary P and the sterol cholesterol affects the growth rates and P stoichiometry of a freshwater keystone species, the generalist herbivorous crustacean Daphnia magna. We predict a decrease in growth rates with decreasing availability of both dietary P and cholesterol and hypothesize a decline in body P content with increasing P limitation, which might be attenuated with increasing cholesterol limitation. Furthermore, we tested how the positive relationship between growth rate and body P content predicted by the GRH within a consumer species depends on the colimitation by cholesterol and set out to detect mechanisms that can lead to potential divergences from the predicted relationship.

Materials and methods


The stock culture of Daphnia magna was raised in filtered water (0·2-μm pore-sized membrane filter) of Lake Stechlin (north-east Germany) with saturating amounts of the P-replete cultured green alga Scenedesmus obliquus (SAG 276-3a, culture collection Goettingen, Germany) as food. For the growth experiment, the well-ingestible, nontoxic and sterol-free cyanobacterium Synechococcus elongatus (SAG 89.79) (Von Elert, Martin-Creuzburg & Le Coz 2003) was used as food for D. magna. S. elongatus (SYN) was cultured in aerated 2-L flasks containing WC medium with vitamins (Guillard 1975) and diluted daily (dilution rate 0·2 day−1) to ensure nutrient and phosphorus (P) repletion. SYN was maintained at an illumination of 40 μmol m−2 s−1 using a 16 h:8 h light/dark cycle. For P-limited S. elongatus (SYN P-), an aliquot of previously P-sufficient cultured SYN was transferred to P-free WC medium and cultivated until molar C/P ratios reached at least 1000. KCl (100 μm) was added to P-free WC medium to prevent a limitation by potassium (K) because of the omission of the K- and P-containing stock solution (50 μm K2HPO4). All organisms were raised at 20 °C.

Experimental design and procedure

To examine the simultaneous dependency of D. magna’s growth and body P content on dietary P and cholesterol, we generated supply gradients of P as well as cholesterol in the food of juvenile animals. Third-clutch juveniles used for the experiment were collected within 12 h from mothers who were previously transferred to P-deficient food (SYN P-) to avoid a temporally different P supply to newly hatched juveniles and thus a potentially confounding variation in P storage. A subset of these juveniles were dried and weighed for the determination of the initial dry mass. Seven P levels and seven different cholesterol concentrations were created, resulting in 49 dietary treatments which were replicated three times (= 147). Replicates started with six juveniles in jars containing 250 mL of food suspension with cholesterol and ten juveniles in jars containing 300 mL of food suspension without cholesterol. Throughout the experiment, daphniids were transferred daily into jars with renewed food suspensions. The experiment was terminated after 7 days, and daphniids of each replicate were split into two groups and transferred separately into preweighed aluminium boats. After drying for 48 h at 50 °C, daphniids were weighed on an electronic balance (±1 μg; CP2P, Sartorius, Goettingen, Germany) and the dry mass of daphniids was determined as the mean from the two aluminium boats. The somatic growth rates were calculated as the change in dry mass per individual from the beginning (DM0) to the end of the experiment (DMt) using the equation


with t as the duration of the experiment in days. After weighing, one of the two samples per replicate was used for the analysis of daphniids’ P content.

Preparation of food

The water of Lake Stechlin contained little dissolved P (6·1 ± 3·6 μg P L−1, mean ± SE, = 3), which may affect the desired dietary P levels of experimental food suspensions (i.e. SYN). Therefore, we added previously P-limited cultured S. obliquus (0·3 mg C L−1) to lake water which depleted the dissolved P concentration in the water as a result of uptake. After 12 h, S. obliquus together with the incorporated P was removed carefully from the water by low-pressure filtration (500 mbar, 0·2-μm pore-sized membrane filter). The P-depleted lake water did not change the C/P ratio of P-limited cultured SYN (SYN P-) after 40 min (two-sample t-test, = 2·7, d.f. = 2, = 0·114). Thus, the P-limited cultured S. obliquus sufficiently reduced the dissolved P in lake water to levels that should not affect the desired C/P levels of SYN.

Only SYN P- was used in the experiment and pulsed with P shortly before preparation of the different C/P levels to prevent potential indirect P effects, such as change in fatty acid composition or different digestibility, caused by different culture conditions (Ravet & Brett 2006). It is assumed that Daphnia is an unselective filter feeder (DeMott 1986) and does not therefore handle individual cells of small size, such as Synechococcus cells. The advantage of the used method (i.e. to use only SYN P- with or without supplementations) is that D. magna was supplied with cells of very equal size, shape and concentration of other biochemical components, making discrimination between the cells of different C/P ratios very unlikely. Before the daily preparation of food suspensions, SYN P- (C/P = 1000) was incubated for 40 min with P (75 μm K2HPO4), providing P-sufficient SYN P* (C/P = 100). An incubation time of 40 min was sufficient for SYN P- to obtain constant low C/P ratios (see Fig. S1 in Supporting Information). Subsequently after P incubation, P concentrations of SYN P- and SYN P* were determined daily as mean of duplicate measurements and SYN P- and SYN P* were incubated with cholesterol. After cholesterol incubation, the desired C/P ratios were achieved by mixing SYN P- (C/P = 1015 ± 18, mean ± 1 SE, duration = 7 days) and SYN P* (C/P = 95 ± 3, mean ± 1 SE, duration = 7 days) in calculated proportions using the previously determined P concentrations. Therefore, the daily mixed C/P ratios varied no more than 5% and did not differ from the desired C/P ratios, which we tested twice during the experiment. To integrate the dietary P concentration over the whole experiment, we calculated the mean of the daily mixed C/P ratios (duration = 7 days).

For the manipulation of dietary cholesterol availability, we incubated SYN P- as well as SYN P* with cholesterol (Sigma, Steinheim, Germany) according to a successfully tested method (Von Elert & Stampfl 2000). For incubation, cholesterol (2·5 mg mL−1) dissolved in ethanol (99·8%; Carl Roth, Karlsruhe, Germany) was added in seven different levels to the sterol-free SYN P- and SYN P* (see Table S1, Supporting information). Excess free cholesterol was removed after four hours of incubation by washing the cyanobacteria twice in fresh P-free WC medium using centrifugation (1560 g for 12 min). Before preparation of food suspensions, aliquots of cholesterol-incubated SYN P- and SYN P* of each incubation level were filtered onto glass fibre filters (Whatman GF/F, 25 mm), transferred to dichloromethane/methanol (2:1, v/v), sonicated for 1 min and stored under nitrogen atmosphere at −20 °C for later analysis of cholesterol amount. The dietary cholesterol concentration of the experimental treatments within each cholesterol incubation level was calculated via the cholesterol concentrations of SYN P- and SYN P* and their mixing ratios for the different C/P ratios. As the cholesterol concentration of SYN P- and SYN P* was different after incubation, it resulted in modest differences in cholesterol gradients among dietary P levels (see Table S1, Supporting information).

The daily prepared food suspensions started with an equal food concentration (2 mg cyanobacterial carbon per litre) which was estimated from photometric light extinction (800 nm) using previously determined carbon-extinction equations. Before the preparation of food suspensions, samples of SYN P- and SYN P* were filtered onto precombusted glass fibre filters (25 mm; GF/F, Whatman) and dried for later analysis of particulate organic carbon using a carbon analyser (HighTOC+N, Elementar, Hanau, Germany) to calculate actual C/P ratios and dietary cholesterol concentrations.

Chemical analysis

We analysed P concentrations for aliquots of SYN P- and SYN P* filtered onto polysulfon filters (25 mm, 0·45 μm; Pall Corporation, Port Washington, NY, USA) and samples of dried daphniids. P concentrations were determined using the molybdate blue reduction method (Murphy & Riley 1962) after dissolving tissues with sulphuric acid and an oxidative hydrolysation using K2S2O8 at 120 °C and 120 kPA by autoclaving.

For the determination of the cholesterol concentration of SYN P- and SYN P* enriched with cholesterol, 5-α-cholestan (Sigma) was added as internal standard to samples (deep frozen in dichloromethane/methanol, 2:1, v/v) and lipids were extracted twice with dichloromethane/methanol (2:1, v/v). Particles were removed by centrifugation (1730 g for 5 min), and the supernatant was evaporated to dryness under nitrogen flow. The evaporated sample was saponified for 60 min at 70 °C with 4 ml 0·2 m methanolic KOH and subsequently extracted three times with isohexane. The neutral lipids were partitioned into isohexane/diethyl ether (9:1, v/v), and this cholesterol-containing fraction was evaporated to dryness under nitrogen and resuspended in a defined volume of isohexane (20–50 μL). Cholesterol was analysed using a gas chromatograph (6890 N; Agilent Technologies, Waldbronn, Germany) with a DB-5 ms capillary column as described previously (Wacker & Martin-Creuzburg 2007). One microlitre of the sample was injected with split 5:1; cholesterol was identified by comparison of the retention times with authentic cholesterol and quantified using the internal standard (5-α-cholestan).

Statistical analysis

For each dietary P level, somatic growth rates (g) of daphniids in response to dietary cholesterol (x) were described by the Von Bertalanffy (1957) growth function


where k is the Von Bertalanffy growth coefficient, gmax is the asymptotic growth rate (i.e. the maximum growth rate at each P level), and g0 is the growth rate without dietary cholesterol. We fixed g0 to 0·063 day−1 as mean growth rate of all P levels without cholesterol supply, because growth rates showed little variation across P levels (SE = 0·002, = 21) and did not differ between those levels (one-way anova, F6, 14 = 1·2, = 0·38). Differences in gmax between the seven dietary P levels were compared on the basis of their 95% confidence intervals (CI) to assess at which P levels growth is limited or saturated by dietary P.

The dependency of Daphnia’s body P content on dietary P as well as cholesterol availability was analysed using analysis of variance (anova) with dietary cholesterol as continuous and dietary P level as factorial variable. Linear regressions at each dietary P level were fitted to test whether and how body P content changes along the dietary cholesterol gradient. The parameters of these regressions (i.e. the regression lines) were used to assess the homoeostatic response of D. magna to changes in dietary P availability. Thereto, we calculated the homoeostasis coefficient H (Sterner & Elser 2002) as the inverse of the slope of the linear regression between the logarithmic fitted values of animal’s P content (fitted body P) versus the logarithmic P content of food (dietary P), both expressed in same units, using the equation


We calculated H continuously along the dietary cholesterol gradient only to 3·0 μg mg C−1 to avoid extrapolation.

To analyse the mechanisms behind the relationship between Daphnia’s body P content and the somatic growth rate as predicted by the GRH applied to the within-species level, we had to consider the dietary manipulations in terms of P as well as in terms of cholesterol. Therefore, we divided our data set into two parts, one of P-limited and another of P-saturated growth rates, and analysed the dependency of growth rates on body P content and dietary cholesterol availability using an analysis of covariance (ancova) in each of the two data sets. To visualize how the relationship between growth rates and body P contents changed depending on cholesterol availability, we categorized dietary cholesterol into four levels: without cholesterol (0 μg mg C−1), low cholesterol (0·6–1·2 μg mg C−1), medium cholesterol (1·3–2·6 μg mg C−1) and high cholesterol (3·2–4·8 μg mg C−1). Linear regressions between growth rate and body P content were applied for each cholesterol category for both P-limited and P-saturated growth to assess the strength of the relationships. All statistical analyses were carried out using the statistical software package R version 2.5.1 (R Development Core Team 2007).


Somatic growth

Our study showed that the growth of D. magna was limited by dietary phosphorus (P) as well as by dietary cholesterol availability (Fig. 1). The growth rate responses at the different P levels followed saturation curves along the dietary cholesterol gradient (Fig. 1, Table 1), i.e. growth rates increased with increasing cholesterol availability up to a saturation plateau. In treatments without cholesterol supply, dietary P had no effect on Daphnia growth rates (anova, F6, 14 = 1·2, = 0·38), which were very low across all P levels (0·05–0·08 day−1). At high cholesterol availability, growth increased with increasing dietary P and levelled off at low C/P ratios. Using fitted gmax, we found P-saturated growth at molar C/P ratios ≤268 and P-limited growth for higher C/P ratios of 449–1015 (Fig. 1, Table 1). This indicates a TERC/P between C/P ratio of 449 and 268 which includes the often proposed TERC/P of 300.

Figure 1.

 Somatic growth rate of Daphnia magna in response to dietary cholesterol concentration at different dietary P levels (C/P, molar). Regression lines indicate nonlinear regressions using eqn 2. Solid lines indicate P levels (C/P = 1015–449) where fitted maximum growth (gmax) differs significantly from the next upper one, whereas dashed lines indicate P levels (C/P = 268–95) where fitted maximum growth (gmax) did not differ from each other (see Table 1). Error bars denote means ± 1 SE (= 3).

Table 1.   Nonlinear regression results of Daphnia magna’s growth rates in response to dietary cholesterol availability at different dietary P levels (C/P, molar) using eqn 2
C/P ratiogmax (95%-CI)t19-value*k (95%-CI)t19-value*
  1. *C/P 1015: t18.

  2. The C/P ratios were calculated as mean ± 1 SE of the daily mixed C/P ratios (duration = 7 days) using SYN P- (C/P = 1015) and SYN P* (C/P = 95; see Materials and methods for further details). All estimated parameters were highly significant (< 0·005). The same letters in superscript indicate gmax that were not significantly different between C/P ratios because of overlapping 95% confidence intervals (CI) given in parentheses.

1015 ± 180·13 (0·12–0·14)a21·91·36 (0·61–2·12)3·8
682 ± 110·18 (0·16–0·19)b26·41·51 (0·87–2·14)5·0
449 ± 80·24 (0·22–0·26)c24·20·79 (0·58–1·00)7·9
268 ± 50·29 (0·26–0·34)d13·20·48 (0·30–0·65)5·7
177 ± 40·28 (0·24–0·32)d14·70·46 (0·30–0·62)6·0
135 ± 30·31 (0·24–0·38)d9·30·33 (0·17–0·49)4·3
95 ± 30·39 (0·25–0·53)d5·70·23 (0·08–0·38)3·2

Phosphorus stoichiometry

The body P content of D. magna showed a nonlinear relationship with dietary P availability (Fig. 2). At the four lowest C/P ratios (C/P 95–268), D. magna’s P content stayed centred around a mean of about 1·2% and declined to a mean of approximately 0·6% with increasing P deficiency. However, within each dietary P level, there was a pronounced variability of Daphnia’s P content (Fig. 2), which could be explained by dietary cholesterol availability (Fig. 3). We found that, besides the strong impact of dietary P (P: F6, 127 = 120·2, < 0·001), dietary cholesterol also affected animals’ P content (cholesterol: F1, 127 = 7·8, = 0·006). Additionally, the interaction between dietary cholesterol and P (cholesterol × P: F6, 127 = 5·4, P < 0·001) indicated different responses of body P content along dietary cholesterol gradients and across dietary P levels. More detailed analyses using linear regressions revealed a decline in body P content with increasing dietary cholesterol for the three highest C/P ratios (Fig. 3, Table 2). At higher P availability (C/P ≤ 268), body P content remained rather constant along the dietary cholesterol gradient (Fig. 3, Table 2). Consequently, body P content was more strongly affected by dietary P availability at high rather than at low dietary cholesterol, indicating a weaker homoeostatic response at high cholesterol availability where faster growth occurred. This is corroborated by the homoeostasis coefficient H which decreased with increasing cholesterol availability (Fig. 3), indicating also a weaker P depletion of D. magna’s tissues at insufficient dietary cholesterol availability.

Figure 2.

 P content (% of dry mass) of Daphnia magna (= 141) in response to dietary P (molar C/P ratio, log-scale).

Figure 3.

 P content of Daphnia magna grown on dietary cholesterol gradients at different dietary P levels (C/P, molar). The regression lines indicate simple linear regressions. Solid lines indicated P levels (C/P = 1015–449) at which the slope differed significantly from zero, whereas dashed lines indicated P levels (C/P = 268–95) with nonsignificant slopes. Error bars denote means ± 1 SE (= 3). The additional axis shows the homoeostasis coefficient H (note log-scale) calculated as the inverse of the slope of the linear regression between the logarithmic fitted values of animal’s P content (regression lines) versus the logarithmic dietary P levels (see equation 3), both expressed in same units. H was calculated continuously only between 0 and 3 μg cholesterol per mg C to avoid extrapolation. A higher value of H indicated a stronger homoeostasis.

Table 2.   Linear regression results of Daphnia magna’s P content in response to dietary cholesterol availability at different dietary P levels (C/P, molar)
C/P ratioSlope (95%- CI)nP value
  1. The C/P ratios were calculated as mean ± 1 SE of the daily mixed C/P ratios (duration = 7 days) using SYN P- (C/P = 1015) and SYN P* (C/P = 95; see Materials and methods for further details). Slopes and their 95% confidence intervals (CI, in parenthesis) are given in (% P per μg cholesterol mg C−1) × 10³.

1015 ± 18−0·69 (−0·97 to −0·41)190·61< 0·001
682 ± 11−0·63 (−1·03 to −0·23)210·370·004
449 ± 8−0·23 (−0·42 to −0·04)190·270·023
268 ± 5−0·17 (−0·48 to 0·13)210·070·243
177 ± 4−0·05 (−0·37 to 0·26)200·010·732
135 ± 30·12 (−0·15 to 0·39)200·050·354
95 ± 30·15 (−0·06 to 0·36)210·110·143

Relationship between somatic growth and phosphorus content

Correlating body P content and growth rates of D. magna without considering cholesterol availability resulted in a weak but still significant relationship (= 0·26, = 0·0015, = 141). To consider the dietary manipulations in terms of P and cholesterol in the analysis of this relationship, we divided our data set into P-limiting (C/P ≥ 449) and P-saturating (C/P ≤ 268) growth conditions (Fig. 4). Under P limitation, growth rates increased significantly with increasing body P content (body P content: F1, 55 = 42·8, < 0·001) after accounting for the growth-improving effect of dietary cholesterol (cholesterol: F1, 55 = 138·8, < 0·001). This positive relationship between body P content and growth rate under P limitation became stronger with increasing cholesterol availability (cholesterol × body P content: F1, 55 = 24·7, P < 0·001) and disappeared in treatments without cholesterol supply (Table 3). Under P-saturating conditions (C/P ≤ 268), after accounting for the primarily growth-enhancing effect of dietary cholesterol (cholesterol: F1, 78 = 775·7, < 0·001), somatic growth showed only a weak relationship with body P content (body P content: F1, 78 = 9·0, = 0·004), which tended to be negative at high cholesterol availability (Fig. 4, Table 3). We found no significant positive correlation between body P content and growth rates under P-saturating growth conditions (Table 3) irrespective of dietary cholesterol availability (cholesterol × body P content: F1, 78 = 0·01, = 0·914).

Figure 4.

 Somatic growth rate (per day) versus P content (% per dry mass) of Daphnia magna (= 141) under P-limiting (filled symbols, C/P ≥ 449) and P-saturating (open symbols, C/P ≤ 268) growth conditions. Growth rates were categorized according to dietary cholesterol availability (circle: without cholesterol, 0 μg mg C−1; triangle: low cholesterol, 0·6–1·2 μg mg C−1; square: medium cholesterol, 1·3–2·6 μg mg C−1; diamond: high cholesterol, 3·2–4·8 μg mg C−1). Regression lines were plotted for each cholesterol category at P-limiting as well as P-saturating growth conditions.

Table 3.   Linear regression results of Daphnia magna’s somatic growth rates (per day) in response to body P content (% P of dry mass) at categorized dietary cholesterol concentrations (μg mg C−1) separately analysed for P-limiting (C/P ≥ 449) and P-saturating (C/P ≤ 268) growth conditions
Dietary cholesterolSlope (95%-CI)nr2P-value
  1. Slopes (per day per % P) with their 95% confidence intervals (CI, in parenthesis) are given.

 P-limiting conditions
00·025 (−0·027 to 0·078)90·160·288
0·6–1·20·110 (0·033 to 0·186)160·400·008
1·3–2·60·122 (0·072 to 0·171)250·53<0·001
3·2–4·80·269 (0·213 to 0·325)90·95<0·001
 P-saturating conditions
00·026 (−0·047 to 0·099)110·070·446
0·6–1·20·035 (−0·020 to 0·090)180·100·200
1·3–2·6−0·046 (−0·119 to 0·027)320·050·209
3·2–4·8−0·109 (−0·255 to 0·038)210·110·137


We showed that the availability of the essential lipid cholesterol as colimiting nutrient strongly influenced the P limitation of D. magna. On the one hand, dietary P had no effect on growth when cholesterol was not available, because all daphniids showed very low growth rates (0·05–0·08 day−1) independent of dietary P availability. On the other hand, the impact of dietary P on growth was high at sufficient cholesterol availability, i.e. growth rates of D. magna ranged from 0·12 to 0·30 day−1 across P levels. This indicates a better utilization efficiency of P for growth when Daphnia was released from strong limitation by another essential food component, such as cholesterol.

The effects of nutritionally imbalanced food are most pronounced at the plant–animal interface and regulate the transfer efficiency of nutrients and energy to higher trophic levels. To predict negative consequences for primary consumers owing to poor food quality, it is important to know at which concentrations nutrients within food become limiting. Threshold elemental ratios, for instance, denote when growth limitation switches from one element to another and are a useful tool to estimate the limitation of consumers by particular elements, such as P (Frost et al. 2006). In our study, we found that the TERC/P for cyanobacterial food enriched with cholesterol was around the previously often observed threshold for Daphnia of 300 (e.g. Plath & Boersma 2001). However, the TERC/P can change with the availability of food, i.e. food quantity (Urabe & Watanabe 1992; Sterner 1997), or with the availability of essential nutrients other than P, i.e. food quality. The latter is supported by our findings that dietary P had no effect on the very low growth rates when the cyanobacterium was not supplemented with cholesterol, which did not allow the determination of a TERC/P. According to this, a potential P limitation of herbivorous zooplankton during cyanobacterial blooms (i.e. high C/P) may become less pronounced because of increasing sterol limitation. Previous laboratory experiments using diet mixtures of the cyanobacterium S. elongatus and phytosterol-containing eukaryotic algae (Scenedesmus obliquus or Nannochloropsis limnetica) showed sterol-limited growth of Daphnia even on mixtures with a high proportion of the sterol-containing algae (Martin-Creuzburg, Wacker & von Elert 2005; Martin-Creuzburg, Sperfeld & Wacker 2009). A proportion of 80%S. obliquus or 50%N. limnetica was necessary to release Daphnia completely from sterol limitation (Martin-Creuzburg, Wacker & von Elert 2005 and Martin-Creuzburg, Sperfeld & Wacker 2009; respectively). Even when not all cyanobacteria can be ingested by Daphnia, these results indicate a high chance of moderate sterol limitation during cyanobacteria dominance. However, owing to the often poor edibility of filamentous or colony-forming cyanobacteria (Wilson, Sarnelle & Tillmanns 2006), an additional carbon/energy limitation may covary with sterol limitation, which would complicate disentangling the effect of these colimiting factors in nature. A sterol limitation of herbivorous crustaceans might not be restricted to the dominance of cyanobacteria, because high light intensity and low P availability decrease the sterol concentration of eukaryotic algae to potentially limiting levels (Piepho, Martin-Creuzburg & Wacker 2010). Furthermore, not all phytosterols of eukaryotic algae are suitable precursors for cholesterol, and other phytosterols are converted, with different efficiency, to cholesterol (Martin-Creuzburg & Von Elert 2004). Thus, the growth of herbivorous crustaceans may depend not only on the amount of sterols in the diet but also on the phytosterol composition (Piepho, Martin-Creuzburg & Wacker 2010), which is determined by the phytoplankton community composition.

The availability of essential nutritional components also has consequences on the animal’s stoichiometry depending on the strength of homoeostatic regulation (Sterner & Elser 2002). Most animals are not strictly homoeostatic in their P stoichiometry (Elser et al. 2003), which is mostly caused by the P availability of the diet, but age can also lead to variations in animal’s P content (DeMott 2003). We found that daphniids’ P content remained constant at P-saturating conditions (C/P ≤ 268), regardless of cholesterol availability (Fig. 3), suggesting the ability to maintain strict P homoeostasis across a wide range of colimiting conditions. Under P-limiting conditions (C/P > 268) however, P homoeostasis was not maintained and body P content decreased with decreasing dietary P, which was also observed in other studies using cladocerans (DeMott, Gulati & Siewertsen 1998; Acharya, Kyle & Elser 2004; DeMott & Pape 2005). Additionally, P stoichiometry of daphniids under P-limiting conditions was influenced by the availability of cholesterol. The decline in D. magna’s P content caused by dietary P deficiency was less pronounced at low compared with high dietary cholesterol indicated by the higher homoeostasis coefficient, H (Fig. 3). The different plasticity of body P content was probably influenced by differences in growth rate capacity because of variations in cholesterol availability. It could be more difficult to maintain P homoeostasis under P deficiency when growth was promoted by high cholesterol availability in comparison with low cholesterol availability when growth rate was low. Thus, P in body tissue might be diluted by rapid growth because of sufficient dietary cholesterol supply. In contrast, when growth rate was low under cholesterol shortage, the amount of P being assimilated was high relative to body mass. DeMott & Pape (2005) investigated the P homoeostasis of ten Daphnia species and found the weakest homoeostatic response for D. magna (lowest H∼ 4), which suggests that D. magna is much more plastic in its P stoichiometry than other Daphnia species. We found that a strong variation in the availability of a particular biochemical, here cholesterol, can even increase the plasticity of D. magna’s P stoichiometry (= 3–6). The variation in the body P content of daphniids probably plays rather a negligible role for consumers of Daphnia (Boersma et al. 2009). Nevertheless, as secondary consumers should be limited at least by food quantity when herbivore growth is limited by food quality, essential biochemicals have to be necessarily considered, besides elements, in studies on the carbon and nutrient transfer of ecosystems (Anderson, Boersma & Raubenheimer 2004).

The effect of biochemicals on animals’ P content found in this study extends the current view that animals’ P stoichiometry is influenced by other food parameters such as the amount of available food (Elser et al. 2003; Ferrão-Filho, Tessier & DeMott 2007; Shimizu & Urabe 2008) or the dietary nitrogen availability (Elser et al. 2003; Acharya, Kyle & Elser 2004). The effects of other limiting food components on animals’ P stoichiometry might also have consequences for the applicability of the GRH within a consumer species. According to the GRH, we expected a positive relationship between growth rate and body P content under P-limiting conditions. However, our results corroborate this prediction of the GRH only partly. We found a positive correlation between growth rate and body P content at sufficient cholesterol availability but not under strong colimitation by cholesterol (Fig. 4). As expected, we did not observe either positive correlations between growth rate and body P content when daphniids were grown under P-saturating conditions. Similarly, other studies did not find the positive relationship between growth and P content within single species (e.g. Acharya, Kyle & Elser 2004; Shimizu & Urabe 2008) when the organisms were grown under non-P-limiting conditions. Within a consumer species, we suggest two possibilities for these non-P-limiting conditions: (i) when the maximum growth capacity is reached, body P content can increase because of increasing dietary P availability and (ii) growth rate responses are largely decoupled from body P content when an essential food component other than P strongly limits growth, which was observed in the treatments without cholesterol supply.

Originally, the GRH focused on between-species differences at P-saturating conditions and predicted that the species with higher specific P content should show higher growth rates. Here, we tested the applicability of the GRH within a consumer species and suggest that the GRH is restricted to P-limiting conditions, when no other nutritional factor strongly colimits growth. This is supported by our observation that under P limitation, the positive relationship between body P content and growth became stronger with increasing cholesterol availability, i.e. with decreasing colimitation by dietary cholesterol. Without cholesterol supply, daphniids showed no growth response with increasing dietary P. Hence, the surplus of P could not be used for growth and indicates that the animals were probably not physiologically P limited. The P content of daphniids varied despite strong cholesterol limitation, which might cause the decoupling of the relationship between growth and body P under limiting conditions of food quality. Recent studies corroborate this prediction because the positive relationship was not found at limiting conditions of food quantity (Elser et al. 2003; Ferrão-Filho, Tessier & DeMott 2007; Shimizu & Urabe 2008) and nitrogen (Elser et al. 2003; Acharya, Kyle & Elser 2004).


The growth and P stoichiometry of our model organism Daphnia were simultaneously affected by dietary P and the availability of cholesterol as colimiting essential food compound. As predicted by the GRH applied to the within-species level, we found positive relationships between growth and body P under P-limiting growth conditions. We conclude that colimitation has a strong impact on the positive relationship when the plasticity in growth and stoichiometry is differently affected by the availability of the colimiting factor, and we emphasize the conditional applicability of the GRH to the within-species level. Our results suggest that the specific body P content is not necessarily a good predictor for potentially P-limiting growth conditions of herbivores in nature and measurements of herbivore’s RNA concentration (Vrede et al. 2004) or alkaline phosphatase activity (Elser et al. 2010) might be more appropriate. We propose that our approach can be used as conceptual framework of how colimiting factors, in general, affect herbivore growth and homoeostasis.


We thank S. Heim for technical assistance. We also thank F. de Castro, William DeMott and an anonymous reviewer for valuable comments on this manuscript and A. Montana for improving the English. This study was supported by the German Research Foundation (DFG, WA 2445/4-1).