Introduction

Allelopathy, i.e. the inhibition of competing organisms through the release of specific chemicals [1], is known to be potentially an important process in aquatic phototrophic microorganisms [2]. It has been shown to influence species successions in lakes [3]. Allelopathic interactions in aquatic microorganisms are under the control of biotic and abiotic environmental factors impacting the production and release of chemicals. For example, allelopathy of the haptophyte Prymnesium parvum is enhanced by unbalanced N/P ratio and by high aeration but is decreased by high light intensity [4]. Allelochemical production by planktonic and benthic green algae is also affected by nutrients and light [5, 6]. Higher temperature, higher light, phosphorous or magnesium deprivation increase the allelopathic activity of the cyanobacterium Cylindrospermopsis raciborskii [7]. Similar responses have been observed in other cyanobacteria [8].

The impact of environmental factors on the production of allelopathic compounds or defence metabolites is generally analysed through classical plant defence theories [9], particularly the growth-differentiation balance hypothesis (GDBH) and the optimal defence theory (ODT). GDBH states that the production of defence metabolites would be enhanced by any environmental factor that reduces growth more than it reduces photosynthesis [10,11,12]. When growth is limited by any stress, the photosynthates that accumulate can be used to produce allelochemicals. Allelopathic potential would therefore be high when environmental conditions (particularly irradiance and CO2 availability) allow high carbon fixation rate by photosynthesis while poor or unbalanced nutrient concentration restricts growth. The results of studies on allelochemical production by freshwater macrophytes were partially consistent with this hypothesis [13, 14].

The main issue with GDBH is that it only considers the cost of defence, but not the benefits. Another of the main plant defence theories, the optimal defence theory (ODT), gives more importance to the needs of a plant for defences [9, 15]. Most studies have considered only one of the consequences of ODT: the different allocation of defence compounds between different parts of a plant [9]; therefore, ODT is not often considered when microorganisms are studied [16]. However, the differential allocation among the plant organs is just one of the ‘subhypothesis’ derived from ODT [17]. Indeed, ODT states that the level of secondary metabolite production is not only controlled by production costs but also by the costs/benefits balance of the production of secondary metabolites [18]. This theory could be extended to allelopathy assuming that the benefits of allelopathy correspond to the decrease in competition and therefore to a better access to resources (light, nutrients and space). The costs correspond to the reduction of growth due to the use of energy, carbon and nutrients to produce and release allelopathic compounds and to potential autoinhibition effects.

In freshwater, allelopathic compounds have very diverse chemical identities, for example fatty acids, peptides, pyrrolidine and alkaloids [19, 20]. Among these, fatty acids or their derivatives are frequent [21]. They are produced for example by the planktonic green alga Chlorella vulgaris [22], by diatom-dominated biofilms [23] and by the filamentous green alga Uronema confervicolum [24]. Besides their role in allelopathic interactions, some of these compounds are part of the primary metabolism of their producers. It is an example of parallel allelopathy where one compound has several identified functions including an allelopathic one [25]. In this prospect, their production is regulated by several factors. For example, changes in light intensity, temperature, salinity and nutrient concentrations as well as a source of organic carbon influence the cellular accumulation of fatty acids in cultured microalgae [26,27,28,29] but also impacts the fatty acids released in the medium [30]. Stream periphyton fatty acid profiles are strongly influenced by light and nutrients [31]. The existence of different regulation systems for the production of multifunction compounds may in some cases lead to non-adaptive concentrations regarding the allelopathic function [25].

The aim of this study was to determine to what extent the regulation of allelopathic compound production in aquatic microorganisms could be related to the need for their allelopathic function, as hypothesised in the ODT, or could be linked to the available resources, as stated in the GDBH. To test these hypotheses, we have used the benthic green alga Uronema confervicolum as a model organism. This species is known to produce linoleic (LA) and α-linolenic acids (LNA), two polyunsaturated fatty acids (C18:2ω6 and C18:3ω3, respectively) with allelopathic properties against benthic epiphytic diatoms and other phototrophic microorganisms [24, 32]. Since LA and LNA are multifunctional compounds, it is necessary to distinguish allelopathy (extracellular activity) from the other functions of fatty acids (intracellular activity).

For this purpose, the two free fatty acids (FFAs) with allelopathic properties (LA and LNA) [24, 32] and three additional FFAs without known allelopathic properties (C16:0 hexadecanoic acid, C18:0 stearic acid and C16:3 hexadecatrienoic acid) have been quantified in the culture medium and in the biomass of U. confervicolum. The alga has been sampled at different culture phases and in various conditions in terms of growth phase, growth rate, nutrient limitation and cell density. ODT predicts that allelopathy will be higher when light or nutrients are low because such situations stimulate the competition for resources. GDBH predicts that the production of allelopathic FFA will be higher in case of nutrient limitation but will also increase with light (because there are more photosynthates available).

Materials and Methods

Algal Cultures

The algal strains used in the study were the green alga U. confervicolum (uro co 01 t strain), isolated in summer 2005 from biofilms sampled in the river Tarn (Southwestern France), and the diatom Gomphonema parvulum strain SAG1032-1, supplied by the SAG (Sammlung von Algenkulturen Gottingen, Germany; this strain was initially isolated in 1950 and kept in culture ever since). Stock cultures of these strains were maintained in 100-mL Erlenmeyer flasks on COMBO medium [33] in a temperature-controlled chamber at 18 °C (light/dark 16:8, 30 μmol m−2 s−1). For the experiments 500-mL Erlenmeyer flasks bubbled with air and filled with 400 mL of COMBO medium were placed in the same conditions after inoculation with 2.5 × 106 cells of U. confervicolum. In these conditions, both strains developed as biofilms in the bottom and on flask walls. These culture conditions are thereafter referred to as standard conditions. G. parvulum was received axenic and kept in axenic conditions while U. confervicolum was not axenic and always kept associated bacteria.

Fatty Acid Adsorption on Biomass

This first experiment was designed to evaluate the potential adsorption of FFA on algal biomass. This was crucial to determine to what extent the fatty acids quantified in the culture medium represented the complete amount of fatty acids released or only a part of them. The adsorption of LA and LNA on a biofilm of the diatom G. parvulum was determined. This diatom was chosen because it contains neither of the FFA used [34]. G. parvulum was cultured axenically in 500-mL flasks filled with 400 mL of COMBO medium renewed regularly by discarding half of the culture medium without resuspending the cells and replacing it by 200 mL of fresh culture medium. After 6 weeks, 350 mL of medium was discarded before the addition of 0.4 mg of LA and 1.3 mg of LNA (purchased from Sigma-Aldrich) in 10 μL of ethanol (corresponding to approximately half of the mean total amount of LA and LNA in a culture flask of U. confervicolum). After 4 h, biomass and medium were separated by filtration (GF/F, 0.7 μm pore size). Biomass was freeze-dried and weighed (± 0.1 mg). FFA concentrations were determined in both filtrate and biomass.

Experimental Design

The concentrations of FFA in U. confervicolum biomass and culture medium have been measured in three different experiments. In experiment A, FFA concentrations were measured throughout U. confervicolum growth in batch cultures in which nutrient concentrations (resources) decreased progressively due to the consumption by the algae and light availability decreased as well because of shading by algal biomass. This was designed to test whether environmental conditions impacted on FFA production and release. Experiment B was designed to test the hypothesis of a regulation of FFA release by the FFA concentration in the culture medium. In experiment C, the crossed effects of carbon and nutrient limitations on FFA concentrations were assessed to determine the important factors.

Experiment A: Variation of FFA Concentrations During Growth

This experiment was designed to study the production and release of FFA during the successive growth phases corresponding to algal growth in batch cultures. U. confervicolum was cultured in standard conditions. The aim was to test the hypothesis of an impact of environmental conditions on fatty acid release. After 4, 7, 10, 12, 14, 17, 20, 24, 31, 38 and 47 days of culture, two randomly chosen flasks were pooled for each of three replicates. Whole algal biomass and medium were harvested and separated by filtration (GF/F, pore size 0.7 μm; multiple filters were used when necessary). The medium was extracted first and then the biomass was scraped using a cell scraper. Algal biomass was weighed after freeze-drying. FFA concentrations were determined in both algal biomass and culture medium, using 800 mL per replicate. For the first date, three flasks were used per replicate in order to have enough biomass for subsequent analysis. Freeze-dried powdered biomasses from the 12, 14, 17, 24 and 47 sampling days were analysed to determine carbon, nitrogen and phosphorous concentrations.

Experiment B: Link Between Algal Density and FFA Concentration in the Medium

This experiment aimed at determining if, when available resources are constant, the production of allelochemicals is constant or if it is regulated so that their concentration in culture medium remains constant. To keep similar light and nutrient conditions but different allelochemical dilution treatments, the same initial biomass of U. confervicolum was inoculated in different volumes of culture medium. If the amount of allelochemicals released by unit of biomass only depended on resources, the concentration of allelochemicals was expected to be lower when the ratio of algal biomass/culture volume was lower (because allelochemicals are diluted). In the experiment, the same number (2.5 × 106 cells) of U. confervicolum was added in either one 1000-mL flask filled with 800 mL, two 500-mL flasks filled with 400 mL or four 250-mL flasks filled with 200 mL (in triplicates), placed in standard conditions. We assumed that at the beginning of exponential growth, algal growth rate was not affected by the volume of culture media. Indeed, considering the low density of algal cell, nutrient limitation was not likely to occur (even in small flasks) in the short time of the experiment and auto-shading also remains at a negligible level. The measurement of biomass was performed to verify that there was not any important difference in biomass. The number of flasks in each condition was adjusted to perform FFA analyses on equivalent culture volumes. After 12 days, biomass and medium were separated as in the growth phase experiment and FFA concentrations were determined in the filtrate.

Experiment C: Variation of FFA Concentrations in Case of C and/or Nutrient Limitation

This experiment was designed to study the effect of resource availability and limitations on FFA production to test directly the different ODT and GDBH hypotheses. In this experiment, we manipulated U. confervicolum growth rate (three levels) and nutrient concentration and limitation (four levels). The different growth rates were obtained by modifying light and aeration conditions. The three levels, thereafter referred to as ‘growth’ conditions, were high light level (60 μmol m−2 s−1) and air bubbling (high growth, HG, 0.0308 day−1), low light (30 μmol m−2 s−1) and air bubbling (intermediate growth, IG, 0.0130 day−1), and low light (30 μmol m−2 s−1) and no air bubbling (low growth, LG, 0.0071 day−1). For each growth condition, the growth rate and sampling date were determined by a preliminary experiment. The low light condition (30 μmol m−2 s−1) was chosen as the light intensity at which our stock cultures are kept and the high light (60 μmol m−2 s−1) as the double of the low light treatment. Algae were cultured in 250-mL flasks filled with 200 mL in high growth and in low growth conditions in COMBO medium. A triplicate was regularly sampled; all the biomass was freeze-dried and weighed. The growth curves obtained (Fig. S1) were used to determine the sampling dates in HG (11 and 23 days) and LG conditions (39 and 78 days), and data from experiment A were used for IG condition (28 and 49 days). Cultures were stopped at mid-exponential growth phase (when half on the maximum biomass is reached) and at the beginning of the stationary phase. To obtain the four nutrient conditions, the culture media used were COMBO medium, nitrogen-depleted COMBO medium (nP, NO3 was reduced by 20%), phosphorous-depleted COMBO medium (Np, phosphorous was reduced by 40%), and nitrogen- and phosphorous-depleted COMBO medium (np, where N was reduced by 20% and P by 40%). For each nutrient condition, nutrient limitation was determined as described in Danger et al. [35]. The algae, inoculated at 1 × 107 cells mL−1, were cultured during 34 days in standard condition in each of the four media described above. During this time interval, the limiting nutrients are expected to have been completely consumed. N, P or both were then added to the media. Only addition of the limiting nutrient can support supplementary growth. Absorbance (680 nm, multi-well plate reader Asys UVM 340, Salzburg, Austria) was measured 8 days after the addition of the different nutrients. This wavelength was chosen as close to the peak of absorbance for this alga. An increase of absorbance after addition of a nutrient, compared to no increase in the control, indicates that this nutrient is the limiting factor. These nutrient limitation assays performed for each nutrient condition indicated that U. confervicolum was N-limited in NP, nP and np treatments and P-limited in Np treatment (Fig. S2). At each sampling date, FFAs were quantified in culture medium and biomass. C, N and P content of algal biomass were determined on all samples from the mid-exponential phase.

Sample Processing

For all experiments, whole algal biomass and medium were harvested and separated by filtration (GF/F, pore size 0.7 μm). When necessary, multiple filters were used. After filtration of the culture, medium fatty acids were immediately extracted from medium (800 mL) using solid-phase extraction (SPE) cartridges (LiChrolut EN 200 mg of sorbent, 3 mL; Merck) conditioned with 2 vol of methanol, washed with 2 vol of water and eluted with 6 mL of 100% methanol. Algal biomass was scrapped from the bottom of the flask with sterile scraper and rinsed with culture medium to harvest all algal biomass. Biomass was weighed after freeze-drying. C and N were measured in freeze-dried powdered biomasses with a ThermoFisher Flash 2000 elemental analyser (ThermoFisher). Total phosphorous was measured after persulphate oxidation in acid condition using ammonium molybdate spectrophotometric method.

Free Fatty Acid Quantification in Algal Biomass and Culture Medium

FFA concentrations were determined in culture medium, using 800 mL per replicate, and in algal biomass. Two micrograms and 10 μg of the internal standard heptadecanoic acid were added to the SPE extract and to algal freeze-dried biomass, respectively. Algal biomass was then extracted with methanol–water–dichloromethane (2.5:2:2.5 mL).

The medium SPE and biomass extracts were directly methylated in boron trifluoride methanol solution 14% (SIGMA, 1 mL) and heptane (1 mL) at room temperature for 5 min. After addition of water (1 mL), fatty acid methyl esters (FAMEs) were extracted with heptane (3 mL), evaporated to dryness and dissolved in ethyl acetate (20 μL). FAMEs (1 μL) were analysed by gas–liquid chromatography on a Clarus 600 Perkin Elmer system using a Famewax RESTEK fused silica capillary column (30 m × 0.32 mm i.d., 0.25 μm film thickness) and a flame ionisation detector. Oven temperature was programmed from 110 to 220 °C at a rate of 2 °C per minute and the carrier gas was hydrogen (0.5 bar). The injector and the detector were at 225 and 245 °C, respectively.

Statistics

The FFA concentrations at each date (growth phase experiment) and for each treatment (density effect experiment) were compared statistically using one-way ANOVA and a Tukey’s HSD post hoc test for multiple comparisons. Comparisons of the global FFA composition (by using the proportion of all FFA in the mix, expressed as percent of total FFA) in culture medium and biomass were achieved using a multivariate analysis of variance (MANOVA). Since the data does not meet parametric assumptions, PERMANOVA was used. Three-way ANOVA (with growth phase, growth conditions and nutrient levels as factors) were used for total FFA content and each FFA proportion comparison. Tukey’s HSD post hoc test was used for multiple comparisons. Data were transformed in order to fit to normality and homoscedasticity. When these conditions could not be achieved, the Kruskal–Wallis test was used instead of ANOVA. Correlation coefficients were calculated and tested between elemental composition and FFA concentrations. For all tests, significance threshold was set at α = 0.05. All analyses were carried out using the statistical software ‘R’ (www.R-project.org).

Results

The most abundant fatty acids produced and released by U. confervicolum were LNA, LA, C16:0, C18:0 and C16:3 (Tables S1 and S2) representing respectively (in standard conditions after 24 days) 21.7%, 60.5%, 10.8%, 2.5% and 4.5% of FFA in biomass (total concentration 42.01 μg mg−1) and 3.2%, 6.1%, 50.9%, 35.9% and 3.9% of FFA in culture medium (total concentration 10.51 μg L−1).

Fatty Acid Adsorption on Biomass

The different volumes of the flask had no significant impact on growth (p = 0.117). Neither LA nor LNA were detected in the flask containing G. parvulum biofilm without addition of fatty acid. This confirms the absence of production of these PUFA by G. parvulum. A total of 231 μg of LNA and 60 μg of LA were recovered after SPE from the medium of biofilm-free flasks and 220 and 123 μg of LNA and LA were recovered from flasks with G. parvulum biofilm. For the latter, only 1.04% (± 0.21%) of these fatty acids were extracted from the medium; the other 98.96% were extracted from the diatom biomass.

Growth Phase and Density Effect on Allelochemical Production and Release

During the growth of U. confervicolum, LA concentration in the culture medium remained nearly constant from the first sampling date (4 days) to 38 days (Fig. 1a). Between 38 and 47 days, a significant threefold increase (from 1.36 to 4.45 μg L−1) was observed (p < 0.001). LA content in algal biomass increased with culture age (p = 0.0014): it was low before 12 days and higher after this date. The highest concentration was obtained at 47 days with 5.14 μg mg−1 of dry algal biomass. A high level of LA content was also reached at 14 days (3.71 μg mg−1). Considering the 4- to 38-day period, LA contents in the medium and the biomass were not significantly correlated (p = 0.19). The highest medium concentration of LNA (2.70 μg L−1) was reached after 14 days of cultivation and the lowest after 31 days (Fig. 1b). Medium LNA concentration was not significantly correlated with culture age (p = 0.23) or algal biomass (p = 0.21). LNA content in algal biomass peaked after 14 days as observed in the medium. LNA contents in the medium and the biomass were not significantly correlated (p = 0.18). LA and LNA contents in the biomass were highly correlated (p < 0.001), whereas for the concentrations in the medium no correlation was found (p = 0.51). C16:3 was not detected in this experiment.

Fig. 1
figure 1

Concentration of linoleic (a) and α-linolenic acid (b), hexadecanoic acid (c) and stearic acid (d) in U. confervicolum biomass (open circles) and medium (closed squares). Dashed grey lines represent the trend of biomass accumulation. Vertical bars represent standard error (n = 3). Letters indicate statistically homogeneous groups (p < 0.05)

Full size image

The medium concentration of both saturated fatty acids studied (C16:0, C18:0) increased with culture age (p < 0.001 for both), with a non-significant decrease between 24 and 31 days, followed by a rapid increase (Fig. 1c, d). Their biomass content peaked after 7 days, rapidly decreased between 7 and 10 days and then stayed at a relatively low level.

The quantities of dissolved allelopathic FFA per biomass unit were very high at the first sampling time (4 days) and strongly decreased until 31 days of cultivation (Fig. 2). By day 31, this ratio increased again as algal growth slowed down and the concentration of both allelopathic FFA in the medium increased.

Fig. 2
figure 2

Mean amount of linoleic acid (LA, open squares) and α-linolenic acid (LNA, closed squares) in U. confervicolum medium per unit of algal biomass and in U. confervicolum biomass (grey triangles). Vertical bars represent standard error (n = 3)

Full size image

In experiment B, flasks of different volumes (200, 400 and 800 mL) were inoculated with the same number of U. confervicolum cells. During the short cultivation period, no significant effect of algal biomass was observed (p = 0.117, Fig. S3). The concentrations of both allelopathic FFA in medium from flasks of different volume were not significantly different (p = 0.21 and 0.23) (data not shown). The amounts of LA and LNA per unit of biomass were significantly affected by the flask volume (p = 0.042 and 0.0093 for LNA and LA per biomass unit) (Fig. 3). The quantities of LA and LNA in each flask were strongly related to the culture volume (p < 0.001 for both). The low amount of biomass reached and the low culturing time exclude an effect due to light or nutrient availability.

Fig. 3
figure 3

Mean medium amount of linoleic acid (LA) and α-linolenic acid (LNA) per unit of U. confervicolum biomass after 12 days of growth in 200, 400 or 800 mL of COMBO medium. Vertical bars represent standard error (n = 3). Letters indicate statistically homogeneous groups

Full size image

Effect of Carbon and Nutrient Limitation

In experiment C, FFA concentrations in medium and biomass were determined at mid-exponential phase and at the beginning of stationary phase in three different ‘growth conditions’ and four different culture media. The highest variations in FFA concentrations were observed in algal biomass (Table 1). Growth phase affected total FFA concentration in the culture medium, as it decreased from 45.0 ± 0.2 μg L−1 to 32.0 ± 0.1 μg L−1 between the mid-exponential phase sampling and early stationary phase sampling (p < 0.001, Table S1). The same pattern was observed with FFA in biomass. The FFA compositions were affected neither in medium nor in biomass. However, the part of C18:0 in the excreted FFA and C16:3 in the biomass increased between the two phases.

Table 1 Probability values from PERMANOVA (for compositions) with the proportion of each FFA and three-way ANOVA (for the others) on total amount of FFA, and the proportion of each FFA in U. confervicolum culture medium and biomass
Full size table

Nutrient condition affected only C18:0 among excreted FFA (p = 0.03). C18:0 was higher in np and lower in NP treatment. In contrast, FFA composition, total FFA and C16:0 concentrations in the biomass were affected by nutrient condition (Table 1).

Growth conditions did not affect total FFA concentration in the medium. The fatty acid composition of algal exudates did not change significantly due to the three factors studied, but the interaction between growth and nutrient conditions was very significant (p = 0.001). In culture medium, growth conditions affected only LA and LNA concentrations, which were significantly higher in high growth conditions (Fig. 4). In biomass, growth conditions strongly impacted FFA composition and total amounts of each FFA (Table 1). LA was higher in IG, LNA and C16:0 were higher in LG and lower in IG, and C18:0 and C16:3 were higher in HG (Table S2).

Fig. 4
figure 4

Concentration of linoleic acid (a) and α-linolenic acid (b) in U. confervicolum culture medium with different growth conditions. HG, high growth (high light and aeration); IG, intermediate growth (low light, high aeration); LG, low growth (low light and low aeration). Vertical bars represent standard error (n = 3). Letters indicate statistically homogeneous groups

Full size image

For the two allelopathic FFAs studied and C16:3, the concentrations in the medium were not correlated (or negatively correlated for LA) with the concentrations in the biomass, while for C16:0 and C18:0, the medium and the biomass concentrations were positively correlated in the mid-exponential phase (Table 2). A PCA (Fig. 5) highlights that samples from different growth phases and growth conditions were quite well separated on the basis of their FFA concentrations in the medium and in the biomass. LA, LNA and C16:3 concentrations in the medium were co-linear, and they were independent of concentration of all FFAs in the biomass. Medium concentrations of LA and LNA were correlated with C16:3 medium concentration in the late exponential phase but were not related with the other variables (Table 2, Fig. 5).

Table 2 Probability values of correlations between the different FFA concentrations in the medium and in U. confervicolum biomass
Full size table
Fig. 5
figure 5

Variable and sample ordination after principal component analysis. Arrows represent variables. Variable in brackets are those measured in U. confervicolum biomass; the others were measured in the medium. Variables measured are LA: linoleic acid, LNA: α-linolenic acid, C16: hexadecanoic acid, C18: stearic acid and C16:3: hexadecatrienoic acid. Symbols represent each sample. Circles: NP treatment; squares: nP treatment; diamonds: Np treatment; triangles: np treatment; open symbols: mid-exponential growth sampling; closed symbols: late/end of exponential growth sampling. Black symbols: ‘high growth’ treatment; grey symbols: ‘intermediate growth’ treatment; light grey symbols: ‘low growth’ treatment

Full size image

C, N and P composition of algal biomass were determined for all treatments at the mid-exponential phase sampling (Fig. S4). The total concentration of fatty acids in the medium was not correlated with algal elemental composition (C, N, P, Tables 3 and 4). N content influenced the composition of excreted FFA (p = 0.031) by lowering the concentration of C18 (p = 0.005). The concentration of fatty acids in algal biomass was correlated with C and N contents (p = 0.001 and 0.026) and negatively correlated with C/N ratio (but p = 0.051).

Table 3 Correlation coefficients of total fatty acid and the five individual FFAs on U. confervicolum biomass C, N and P content and C/N, C/P and N/P ratio in algal biomass
Full size table
Table 4 Correlation coefficients of linear regression of total fatty acid and the five individual FFA on U. confervicolum biomass C, N and P content and C/N, C/P and N/P ratio in algal biomass
Full size table

Discussion

Distinguishing Allelopathy from Other Functions of Allelochemicals

The involvement of FFA in U. confervicolum allelopathy has been shown in a previous study [24]. In such a study, a doubt remained due to the low FFA concentration found in the culture medium. Here, our results show that, despite a low concentration in the culture medium, FFAs are highly concentrated in the vicinity of biofilm due to the adsorption of hydrophobic compounds. In fact, nearly 99% of the fatty acids added in the medium were found adsorbed on diatom biomass. This confirms that FFAs produced by U. confervicolum can act as allelochemicals against diatoms in biofilms since they reach their target in sufficiently high quantity. Allelopathy mediated by FFA is likely a case of parallel allelopathy, i.e. allelopathy is not the unique function of the allelochemical [25]. FFAs are involved in both primary metabolisms and allelopathy [32]. It is common that metabolites have multiple functions (reviewed by Pohnert [36]). For instances, brevetoxins produced by the dinoflagellate Karenia brevis are involved in grazer deterrence and are also a stock of carbon and nitrogen [37]. Haemolytic compounds from the Raphidophyceae Fibrocapsa japonica have been found to be simple FFAs [38]. In the case of parallel allelopathy, it could be expected that the regulation of allelochemicals will not always be adaptive considering the allelopathic function. Measuring excreted FFA is the only way to evaluate the regulation of the allelopathic function mediated by these compounds. Indeed, by definition, allelochemicals have to be externalised to be effective. Excreted compounds could be adsorbed on surfaces, including algal biomass [39]. Our ‘adsorption experiment’ showed that a high quantity of FFA could be adsorbed on a diatom biofilm. A diatom was chosen here because this species does not produce LA and LNA ([34] and confirmed by our results), but the different cell surface properties and the surface/volume ratio of the diatom cells could have increased adsorption compared with U. confervicolum filaments. It is likely that FFA measured in culture medium accounts for only a minor part of the extracellular FFA of U. confervicolum cultures. Most of it is certainly adsorbed on the cell surface. This indicates that we should consider the extracellular pool of FFA as divided into medium FFA and adsorbed FFA. Adsorbed FFA and intracellular FFA were not distinguishable. In our experiments, medium FFA concentration could be linked to the extracellular FFA/biomass ratio more than to the extracellular FFA amount. In any case, an increase in medium concentration of a FFA would be due to an increase in FFA release per algal cell. The importance of allelochemical adsorption has been shown with fischerellins produced by the cyanobacterium Fischerella muscicola [40]. It is also quite similar to what is found with macroalgae expressing surface-associated allelochemicals [41, 42]. Adsorption of allelochemicals is rarely taken into account. The production of allelochemicals is often presented as a unit of compounds quantified in the medium per unit of algal biomass (e.g. [22]). However, when the compound is strongly adsorbed by algal biomass, the measure already depends on algal biomass.

The externalisation of allelopathic compounds could be the result of passive mechanisms: passive diffusion or damaged cells [23], or the result of active exudation controlled by the cell. In the case of passive mechanisms, compound concentration in the medium would highly correlate with algal density. Moreover, in the case of passive diffusion, the concentrations of FFA in both biomass and culture medium would have been positively correlated. The results of the experiment C (abiotic factor experiment) revealed that the concentrations of the two allelochemicals LA and LNA in the biomass were negatively correlated or not correlated with their concentration in the medium (Table 2, Fig. 5). Moreover, the amount of LA and LNA remained relatively constant while biomass increased in the experiment A ‘growth phase experiment’ (Fig. 1). These results indicate that there are two pools of these FFA, the extracellular ‘allelopathic’ pool and the intracellular pool, with specific regulation mechanisms. Other studies showed that the FFA released in the medium can also inhibit the development of the producer [30]; therefore, allelopathy may be only one of the extracellular FFA functions and the autoinhibition effect may also impact the regulation of FFA release.

To our knowledge, C16:3 has never been reported as an allelopathic compound, but, as for the two allelopathic FFA, the medium concentration of C16:3 was not correlated with biomass C16:3 concentration (Fig. 5, Table 2). Nevertheless, an allelopathic function could be expected, as the cytotoxic effect of FFA on algae is generally high with highly unsaturated FFA [43]. C16:0 and C18:0 have not been identified as U. confervicolum allelochemicals.

Regulation of the Release of Allelopathic Compounds

Growth differentiation balance hypothesis (GDBH) states that the production of carbon-based secondary metabolites increases when growth is limited and excess of photosynthates is available for the synthesis of secondary metabolites [11]. Optimal defence theory (ODT) does not take into account only the cost of defence (or allelopathy) but also the benefits of the investment in such compounds [17, 18]. GDBH assumes that priority is always given to growth, while ODT postulates that resources could be allocated to secondary functions if the benefits of these secondary functions are higher than those of an allocation to growth. FFA could be easily induced because they can be stored in triglyceride and rapidly released [36]. In our study, we observed a strong increase in medium LA concentration after 38 days of growth and a less strong increase for LNA between 31 and 47 days (Fig. 1b), corresponding to high C/N and C/P ratios in algal biomass (Fig. 6) as it is predicted by the GDBH. However, a peak in LNA concentration (and for LA, though to a lesser extent) was observed around the 14th day of the experiment (Fig. 1a, b). We suppose that this peak began at the time when nutrients were no longer available in the medium, according to previous results about growth and nutrient concentration in the same culture condition (Leflaive and Ten-Hage 2009). Therefore, stored nutrients began to be used by the alga (increase of C/N and C/P ratio after this date, Fig. 6). The alga could have released this allelopathic compound to inhibit competitors at this ‘critical’ point. At a certain nutrient level, carbon could be allocated to allelopathic compounds in order to limit the depletion of nutrients by other species. This hypothesis is consistent with the ODT since allelopathic FFA is expected to increase with the need of allelopathic compounds to exclude competitors.

Fig. 6
figure 6

C/N (a) and C/P (b) ratios in U. confervicolum biomass. Algae were cultured in standard conditions. Vertical bars represent standard error (n = 3). Letters indicate statistically homogeneous groups

Full size image

The stability of the allelopathic FFA concentration in the culture medium, while biomass increased, besides the phenomenon of absorption, could imply a regulation process. Experiment B was designed to test the hypothesis of a regulation of FFA exudation through a negative feedback process. The cultivation time was sufficiently short to keep U. confervicolum in exponential growth with a similar growth rate in each treatment (Fig. S3). In this experiment, a lack of regulation of the exudation would have resulted in a constant FFA quantity (and different concentrations) because there were no significant differences in algal biomass. According to GDBH, the amount of FFA excreted in the medium should have been the same whatever the volume of the flask containing the algae because the culture conditions were identical (at least at the beginning of exponential phase). For the same reason, according to the benefits-based approach (ODT), the concentration of FFA should instead remain the same because the need for effective allelochemicals was constant. This implies the existence of a negative feedback control of FFA exudation. Our results are consistent with this latter hypothesis (Fig. 3).

GDBH predicts that the production of carbon-based allelochemicals will be the highest when conditions for photosynthesis allowed for maximal carbon fixation (our HG condition). Conversely, when using ODT to predict about allelopathy, an increase in allelochemicals in low light conditions (mimicking shading by planktonic or epiphytic competitors) could be expected. In our ‘abiotic factor experiment’ (experiment C), the highest concentrations in allelopathic FFA were reached in HG conditions (Fig. 4, Table 1), which is consistent with the GDBH and some previous results. Indeed, increase of allelopathy with increased light intensity has been observed with the cyanobacterium Cylindrospermopsis raciborskii [7]. On the contrary, allelopathy of the haptophyte Prymnesium parvum decreases when light intensity is increased [4] and allelopathy of the green alga Chlamydomonas reinhardtii has been found to be higher with reduced light [5]. Results obtained by Gross [14] on allelopathic phenolic compounds exuded by Myriophyllum spicatum were more contrasting. Indeed, increasing light levels increased the amount of exuded phenolic compounds (supporting GDBH), but the proportion of phenolic compounds in the total organic carbon exuded became higher when light was reduced [14]. Therefore, the allocation of carbon to allelopathy decreased with decreasing light, but less than the allocation to other exuded compounds. This was consistent with the ODT. However, from the effects of light reported in the literature, some results can also be due to saturating light level and photoinhibition. The different light levels used here were named ‘high’ and ‘low’ relatively to each other and could not be taken as an absolute scale. This makes the comparison of light effect among different studies a difficult task.

Light limitation and CO2 limitation both reduce growth rate (Fig. S1). However, according to Van de Waal et al. (2009) [44], growth limitation by light or by CO2 availability has different effects on secondary metabolites production. In our study, we did not directly assess whether light or CO2 availability limited growth in each treatment. However, since additional aeration without additional light allowed a higher growth in IG treatments than in LG (Fig. S1), we could assume that growth rate was limited by CO2 in LG treatments. Similarly, since additional light without additional aeration allows a higher growth in HG condition than in IG (Fig. S1), we could assume that growth rate was limited by light in IG condition. The highest quantities of LA, LNA, C16:0 and C18:0 in algal biomass at the beginning of the stationary phase were obtained in the intermediate condition where the CO2 availability/light ratio was the highest (Table 1). This highlights the fact that CO2/light ratio is a factor to consider when studying regulation of allelopathy. In our case, for both allelopathic FFA, this amount of excreted FFA followed an increasing trend from LG to HG. Therefore, allelochemical release depended more on the growth rate (resulting from light and CO2 levels) than on the CO2/light ratio.

Nutrient limitation reduces growth more than it reduces carbon assimilation by photosynthesis and therefore leads to an increase in carbon/nutrient ratio. According to GDBH, an increased production of allelochemicals is therefore expected when cells are nutrient-limited. The effect of nutrient limitation on the benefits of the release of allelochemicals is difficult to predict. Actually, nutrient limitation could increase the benefits of inhibiting competitors because it would limit depletion of nutrients by other species. But target organisms are generally more sensitive to allelochemicals under nutrient-sufficient conditions than under nutrient stress (Leflaive and Ten-Hage 2009), so the benefits of allelochemicals would be higher in nutrient-rich conditions. Our results on the effect of nutrients are not consistent with any of these theories. Actually, LA and LNA concentrations in HG condition were the highest in the treatment with the highest nutrient level (NP) and were much lower in the unbalanced Np and nP treatments (Table S1 and S2). These results confirm the reports obtained on the same species and another green alga [6], but are contradictory to numerous results. Indeed, higher allelopathic activity under P- or N-depleted conditions was observed with different microalgae [4, 7, 8]. Particularly, limiting P concentrations has often been pointed out to increase toxicity [45] and allelopathy [46]. DellaGreca et al. [22] found an increased per cell production of chlorellin (an allelopathic mixture of FFA) in P-limited growth. Here, the different nutrient levels have not affected the release of allelopathic FFA (Table 1). It could be due to low differences between nutrient levels (20 and 40% lower N or P, respectively). Many studies used larger nutrient-level differences [6, 22]. However, lower nutrient levels would have led to very a different final amount of biomass and therefore to incomparable conditions.

We measured C, N and P cellular concentrations in order to determine whether allelopathic FFA were affected by biomass elemental composition (Fig. 6, S4). Indeed, Pinna et al. [37] have shown that C-rich toxin production by the marine dinoflagellate Ostreopsis cf. ovata was positively correlated with high cellular C/N and C/P ratio. In culture medium, only C18:0 was affected by N content and C/N ratio and C16:3 was positively affected by N/P ratio (Tables 3 and 4). N and C contents had strong effects on biomass FFA concentrations for all FFA (Tables 3 and 4). Algal growth would depend on protein synthesis; therefore, it depends on N availability and on the rate of protein synthesis that depends directly on P availability (as the majority of P is used to make ribosomes). The assimilation of carbon depends on proteins involved in photosynthesis. Therefore, N or P depletion limits the assimilation of carbon and its allocation to secondary metabolisms [47]. The measurements of biomass C, N and P content do not take into account the loss of these elements through excretion of compounds, but this may be negligible compared to the biomass content.

Conclusion

This study is the first to chemically quantify the effects of abiotic factors on the production of allelopathic compounds by benthic filamentous green algae. Growth was the major factor, affecting positively the release of allelopathic FFA. Both GDBH and ODT were insufficient to explain the observed patterns of regulation (Fig. 7). The results of some or our experiments could be explained by either theory, but neither of them can explain all the results. However, benefits of allelopathy are rather hard to estimate; the evaluation of benefits still needs to be refined to completely test the hypotheses. We focused on two allelopathic FFA, but U. confervicolum also produces other unidentified compounds with allelopathic activity [24]. It is possible that resources are differentially allocated to different kinds of metabolites with a secondary function, depending on environmental factors. Such trade-offs have been recently highlighted in marine seaweeds exposed to competitors [48]. Moreover, in a complex environment, the diversity of interactions (competition with other species and other trophic interactions) could have led to non-adaptive processes regarding interactions between U. confervicolum and diatoms. Our results highlight the limits of some ecological theories that are well suited to simple systems but not to complex interactions with many feedback mechanisms and allocation trade-offs. Further research is needed because elucidating the regulation of allelopathic compound production could be a way to access how these interactions are important and how much organisms could invest in these interactions.

Fig. 7
figure 7

Schematic representation of the effect of different factors on allelochemicals according to growth-differentiation balance hypothesis (GDBH), optimal defence theory (ODT) and the results of the present study

Full size image