Algal and cyanobacterial secondary metabolites in freshwaters: a comparison of allelopathic compounds and toxins

Authors


Joséphine Leflaive, Laboratoire d'Ecologie des Hydrosystèmes, UMR CNRS 5177, Université Paul Sabatier, 118 route de Narbonne, 31062 Toulouse Cedex 9, France. E-mail: leflaive@cict.fr

Summary

1. The photoautotrophic micro-organisms collectively termed ‘micro-algae’ (including micro-eukaryotes and cyanobacteria) are known to produce a wide range of secondary metabolites with various biological actions. A small subset of these compounds has been identified. Some of them, termed allelopathic compounds, have been shown to play a role in allelopathy, defined here as inhibitory effects of secondary metabolites against either competitors or predators. Freshwater cyanobacteria also produce some secondary metabolites, termed toxins, which are highly toxic for animals.

2. While allelopathic compounds play a role in the interactions between the emitter organisms and their direct competitors or predators, toxins are categorised according to their toxic effect on several organisms, including some that may not be present in their immediate environment. However, these two definitions are not mutually exclusive. This review considers the evolutionary, ecological and physiological aspects of the production of allelopathic compounds by micro-algae in freshwaters, and compares the characteristics of allelopathic compounds with those of toxins.

3. Allelopathic compounds include alkaloids, cyclic peptides, terpens and volatile organic compounds. Toxins include alkaloids, cyclic peptides and lipopolysaccharides. No allelopathic compound type is associated with a particular phylogenetic group of algae. In contrast, freshwater toxins are only produced by cyanobacteria belonging to a restricted number of genera. Allelopathic compounds have various modes of action, from inhibition of photosynthesis to oxidative stress or cellular paralysis. Toxins are often enzyme inhibitors, or interfere with cell membrane receptors.

4. The ecological roles of allelopathic compounds have been well identified in several cases, but those of toxins are still debated. In the light of descriptions of negative effects of toxins on both micro-invertebrates and photoautotrophic organisms, we suggest that at least some toxins should actually be considered as allelopathic compounds. Further research on toxic secondary metabolites in freshwaters is now needed, with emphasis on the ecological effects of the compounds in the immediate environment of the emitter algae.

Introduction

The term ‘allelopathy’, from the Greek word ‘allelon’ meaning mutual and ‘pathos’ meaning harm or affection, was introduced by Molisch (1937) to designate the process by which one plant influences another by chemical means. Rice (1984) has included micro-organisms (bacteria, fungi and micro-algae) in this definition and considered both positive and negative effects on the target organism. By analogy with plant–insect interactions, predator defences are sometimes included in the definition of allelopathy (Rizvi & Rizvi, 1992). Because of the complexity of the interactions in natural ecosystems, definitive evidence for allelopathy in the field is almost impossible to obtain. Nevertheless, many field and laboratory studies have pointed to the existence of allelopathic interactions, especially due to secondary metabolites produced by micro-algae (see reviews by Maestrini & Bonin, 1981; Lewis, 1986).

For convenience here we follow the traditional practice of grouping both cyanobacteria and photoautotrophic micro-eukaryotes under the term ‘micro-algae’ even though these two groups are phylogenetically quite distinct. Indeed, we emphasise ecological aspects and in that context no distinction is needed between cyanobacteria and photoautotrophic micro-eukaryotes. Freshwater algae, like marine algae, are known to produce a wide range of secondary metabolites (Table 1). These diverse compounds are released into the environment during algal growth or at cell lysis, but for many no biological or physiological activity has been ascribed to date. Those compounds that affect in a positive or negative way other organisms and hence affect the structure of ecosystems are termed allelochemicals. Among allelochemicals, some compounds that have been shown to have anti-algal, antibiotic, antifungal and anti-predator activities are allelopathic compounds (Smith & Doan, 1999; Gross, 2003; Legrand et al., 2003). Allelopathic activity is a widespread phenomenon amongst freshwater primary producers (Inderjit & Dakshini, 1994). The production of allelopathic compounds is highly species-, and even strain-dependent. Few compounds have been chemically identified to date, despite the increasing number of allelopathic interactions described between micro-algae. In lakes, allelopathy is suspected to play an important role in the establishment of algal successions and in the formation and ending of blooms (Keating, 1977; Vardi et al., 2002; Takamo et al., 2003). In rivers and streams, where algal exudates may be rapidly carried away by the current, allelopathy may be less important in planktonic communities, although it can still be a factor in benthic communities. Some allelochemicals which present acute toxicity against animals (Jochimsen et al., 1998; Griffiths & Saker, 2003) are grouped under the term ‘toxin’. In freshwaters these are mainly produced by cyanobacteria (Carmichael, 1997). Because they represent a health hazard, increasing with the eutrophication of inland waters which facilitates the formation of dense blooms of toxic cyanobacteria, these secondary metabolites have been extensively studied during the last 20 years.

Table 1.   Main algal groups and the chemical natures of associated toxic and allelopathic secondary metabolites produced in freshwater (bold) and marine environments
Taxonomic groupsPhycotoxinsAllelopathic compounds
  1. LPS, lipopolysaccharides.

DinophytesAlkaloids, polyethersFatty acids, polyethers, unidentified compounds
Euglenophytes
Green algaeAlkaloids, estersFatty acids, unidentified compounds
HaptophytesPolyethersPolyunsaturated aldehydes, polyethers
Ochrophytes
 Bacillariophyceae (diatoms)Domoic acidPolyunsaturated fatty acids, polyunsaturated aldehydes
 PhaeophyceaeFatty acid derivates
 RaphidophyceaeFatty acids? polyethers?Aldehydes
RhodophytesDomoic acidUnidentified compounds
Cyanobacteria
 ChroococcalesCyclic peptides, alkaloids, LPSPeptides
 NostocalesCyclic peptides, alkaloids, LPSCyanobacterin, nostocyclamide, nostocyclamide M, nostocine A
 OscillatorialesCyclic peptides, alkaloids, LPS
 StigonematalesLPSFischerellin A and B, alkaloids

This review considers the evolutionary, ecological and physiological aspects of the production of allelopathic compounds by micro-algae (including cyanobacteria) in freshwater environments, and compares the characteristics of allelopathic compounds with those of toxins. In the light of recent studies on the pattern of secreted cyanobacterial peptides (Welker, Christiansen & van Döhren, 2004), toxins may appear as a part of a more important group of bioactive compounds comprising both toxic and non toxic compounds. We argue that some freshwater algal toxins should be considered as allelochemicals active against both competitors and predators, i.e. allelopathic compounds (Fig. 1).

Figure 1.

 Examples of interactions involving allelochemicals (allelopathic compounds and toxins) produced by freshwaters micro-algae.

Conceptual and methodological aspects of allelopathy

Evolutionary aspects

Allelopathy may be either the result of a direct selection of secondary metabolism, or a secondary process where the biosynthesis of molecules was originally selected for other purposes (Reigosa, Sanchez-Moreiras & Gonzales, 1999). It may have developed when the emitter organism first released some compounds in order to avoid their autotoxicity or when mechanisms of self-resistance evolved, which could then have led to a secondary advantage. In the case of terrestrial plants, allelopathy may have served primarily to protect the plant against attack by fungi or micro-organisms. For those organisms, the synthesis of defence metabolites is constitutive or inducible (Tang et al., 1995). Most authors adopt the view that allelopathy originated as a byproduct of other ecological processes.

A major issue in evaluating the impact of allelopathy is that for long-term co-existence in the same habitat, organisms are necessarily adapted to each other. This implies that allelopathic interactions are transitory and in most cases not apparent because of co-evolution (Reigosa et al., 1999). Allelopathy should become apparent in cases of abiotic stress, invasion by exotic organisms, synthesis of a new molecule by the emitter organism, delay in the target adaptation, or accumulation of allelopathic compounds in the environment. Stress can enhance both the production of allelopathic compounds and the susceptibility of the target. In spite of adaptations, Legrand et al. (2003) considered that allelopathic interactions should be widespread in aquatic environments. Natural selection should favour allelopathic compound production, given that this reduces competition and thus improves resource availability. Some targets may become adapted to an allelopathic compound, but in a complex community with a mix of different species some will remain sensitive to the compound. This can confer a weak but sufficient advantage for the emitter. Costs associated with the production of an allelopathic compound may decrease this advantage. Their existence is suspected but they are still unidentified (Legrand et al., 2003). Several instances have been reported that illustrate co-evolution between aquatic freshwater micro-organisms with respect to chemical interactions; for example, physiological resistance to toxins in the freshwater crustacean Daphnia magna compared with acute sensitivity of zooplankton that do not coexist with toxin-producing cyanobacteria (Kurmayer & Jüttner, 1999). In some cases co-evolution is marked by the existence of reciprocal interactions, two organisms producing some compounds acting each on the other organism (Kearns & Hunter, 2000, 2001; Vardi et al., 2002).

In water, chemical information is transmitted by diffusion and advective lamina flow (Wolfe, 2000). Allelopathic compounds with a small molecular weight are favoured because of their faster diffusion. In aquatic environments the distances between cells are quite important and a major problem in the pelagic environment is dilution of the secreted products. Thus Lewis (1986) assumed that allelopathy is not an evolutionarily stable strategy for phytoplankton. Given the distances between cells, the large number of cells present and the importance of viscous forces at that scale, algae of the same species but also those of other species could benefit from the presence of an allelopathic compound (what Lewis, 1986, called ‘distributed benefits’). These algae avoid costs associated with the production of allelopathic compounds but have all the benefits. Group selection of individuals sharing the same genome is not a satisfactory explanation because, unlike terrestrial plants and benthic algae, phytoplankton lack any fixed spatial association (Lewis, 1986). As a consequence, Lewis (1986) considered only the ‘allelochemical-signal hypothesis’ to be realistic; for the emitter organism, the allelopathic compound is a byproduct of the metabolism, but for the target, it is an indicator of the position of the environment which has an effect on its life cycle. In the light of recent work on the diverse mode of action of allelopathic compounds inside the target cell (see Modes of action), Lewis's theory looks quite unrealistic. One of his postulates, the importance of viscous forces, is now questioned by advances in flow mechanics whereby the environment in the vicinity of algal cells is now considered rather stable at the relevant scale. The problem of ‘distributed benefits’ is thus reduced and the advantages from production of allelopathic compounds appear high enough for it to be selected.

In benthic habitats, the physical constraints are different. Epilithic biofilms in rivers are microbial aggregates formed by an association of both photoautotrophs and heterotrophs (both prokaryotes and eukaryotes) surrounded by a polysaccharide matrix. In such a habitat, cellular distances are shorter or even zero (direct cell contacts). Molecules transferred by direct contact or through the polysaccharide matrix can be more lipophilic than in the water column, which means reduced costs for the emitter cell. Inside the biofilm, in addition to nutrient competition, the photoautotrophic micro-organisms are in competition for space through the access to an anchor zone, to light or to nutrient-enriched zones. Space competition adds a supplementary selective pressure which can lead to the formation of allelopathic interactions. Hence biofilms appear quite favourable environments for the appearance of allelopathy and for its study (Jüttner, 1999).

Methods

Research on allelopathy in aquatic environments is focused on: (1) demonstration of the production of allelopathic compounds by an organism; (2) understanding factors influencing production of compounds; (3) identification and characterisation of the compounds and their pathway of biosynthesis and (4) estimation of the role and importance of allelopathic interactions in the field. These objectives necessitate a wide range of methods, from classical culturing to modern chemical investigations. Use of molecular methods will certainly increase in the near future. One common difficulty, particularly for field studies, is to separate competition from allelopathy. Yet first reports of allelopathic interactions in aquatic environment often came from observations in the field (Akehurst, 1931; Hutchinson, 1944; Keating, 1977).

One of the most widely used methods to identify allelopathic interactions is cross-culturing: a target alga is cultured in a medium enriched with filtrate from the culture of another alga whose allelopathic activity is being investigated. Whatever bioassay is chosen, an important issue is the choice of the indicator or target strain. Species that co-exist in the field should be well-adapted to each other and consequently allelopathy is rarely apparent amongst species belonging to the same community (Reigosa et al., 1999). Allelopathic interactions that are not detectable because of adaptation may become evident under physico-chemical stress. In light-, nutrient-, or space-limited conditions the production of an allelopathic compound may be enhanced while the target may become more sensitive. Legrand et al. (2003) suggested that nutrient depletion effects should be avoided in studying allelopathy. Yet, environmentally realistic, nutrient-depleted conditions for both donor and target strains could yield useful information if the effects of competition are well separated from those of allelopathy. The isolation of allelopathic compounds and the determination of their structure require classical chemical methods such as nuclear magnetic resonance, X-rays, UV and mass spectroscopy, high performance liquid chromatography, gas chromatography/mass spectrometry. These methods are coupled with a bioassay to determine which fraction contains the active compound. The bioassay needs to be sensitive, easy to perform and environmentally relevant. A major difficulty for the isolation of bioactive compounds is that they are often produced in very low amounts, as producing a low amount of a highly active compound is a more cost-effective strategy.

Modes of action

Allelopathic compounds

The mode of action of the compound depends on the nature of the interaction between donor and target organisms, the activity of allelopathic compounds being directed against either competitors or predators. In the context of competition, which is mainly with other photoautotrophic organisms, allelopathic compounds may inhibit photosynthesis, kill the competitor or exclude it from the donor vicinity (settling, paralysis). As a predator defence, allelopathic compounds would be efficient by poisoning grazers or by inducing resistant forms in the other algae. In the field, allelopathic compound modes of action are quite various (Table 2). Some examples are given here to illustrate this variability.

Table 2.   Main identified allelopathic compounds and toxins produced by freshwater algae, their effects on different type of organisms and their mode of action when it is known
CompoundEmitterTargetEffectMode of actionReferences
Allelopathic compounds
 Calothrixine ACalothrix sp.BacteriaGrowth inhibitionInhibition of RNA synthesis and DNA replicationDoan et al. (2000)
 CyanobacterinScytonema hofmanniAquatic and terrestrial plantsGrowth inhibitionPSII inhibitionGleason et al. (1986)
  Micro-algaeGrowth inhibitionPSII inhibitionGleason & Paulson (1984)
 2E,4Z-decadienal and other polyunsaturated aldehydesThalassiosira rotula and other diatomsInvertebratesDecrease in fecundity, decrease in egg hatching success, teratogenic effectsInhibition of cell division: microtubule depolymerisationMiralto et al. (1999)
Ianora, Poulet & Miralto (2003)
 Eicosapentanoic acid‘Diatom biofilms’InvertebratesDeathJüttner (2001)
 Fischerellin AFischerella muscicolaFungiGrowth inhibitionHagmann & Jüttner (1996)
  Micro-algaeGrowth inhibitionPSII inhibitionGross et al. (1991)
  InvertebratesdeathHagmann & Jüttner (1996)
 12-epi-hapalindole EFischerella sp.Bacteria and fungiGrowth inhibitionInhibition of RNA synthesisDoan et al. (2000)
 Nostocine ANostoc spongiaeformeMicro-algae Formation of ROSHirata et al. (2003)
 NostocyclamideNostoc sp.InvertebratesDeathTodorova & Jüttner (1996)
  Micro-algaeGrowth inhibitionTodorova & Jüttner (1996)
Toxins
 Anatoxin-aCyanobacteria belonging to the genera Anabaena, Microcystis and OscillatoriaVertebratesMuscles paralysisbinds to nicotinic acetylcholine receptor 
InvertebratesDecrease in fecundity Gilbert (1994); Claska & Gilbert (1998)
Aquatic plantsGrowth inhibition, reduction in photosynthetic oxygen productionFormation of ROS?Mitrovic et al. (2004)
Micro-algaeCellular paralysisKearns & Hunter (2001)
 CylindrospermopsinAphanizomenon ovalisporum, Cylindrospermopsis raciborskii, Raphidiopsis curvata, Umezakia natansVertebratesHeart failure, specific injuries in different organsinterference with protein/ enzyme synthesisGriffiths & Saker (2003)
Invertebratesreduced fitness and growthNogueira et al. (2004)
Terrestrial plantsGrowth inhibition, metabolism inhibitionVasas et al. (2002)
 MicrocystinsCyanobacteria belonging to the genera Anabaena, Microcystis, Oscillatoria and PlanktothrixVertebratesDiarrhoea, vomiting, piloerection, chronic inflammation of liverInhibition of protein phosphatases 1 and 2ABriand et al. (2003)
InvertebratesDeathWiegand et al. (2002)
Aquatic plantsGrowth inhibition, reduction in photosynthetic oxygen productionPflugmacher (2002, 2004, Mitrovic et al. (2005)
Micro-algaeGrowth inhibition, cell deathFormation of ROSVardi et al. (1999); Hu et al. (2005)
 SaxitoxinsAphanizomenon flos-aquae, Anabaena circinalis, Cylindrospermopsis raciborskii, Lyngbia wollei, Planktothrix sp.VertebratesMuscles paralysis, deathBinds to site 1 of sodium channel of the cell membraneCarmichael (1994)

Inhibition of photosynthesis.  Growth inhibition, and eventually, death by inhibition of photosynthesis is a quite widespread mode of action for cyanobacteria. Cyanobacterial allelopathic compounds are generally soluble in organic solvents, insoluble in water and have a low molecular weight. These properties help them to reach the thylakoid membranes where photosynthesis occurs (Smith & Doan, 1999). Allelopathic compounds produced by the cyanobacteria Scytonema hofmanni (cyanobacterin) and Trichormus doliolum (unidentified compound) both inhibit the photosystem II-mediated photosynthetic electron transfer (Gleason & Baxa, 1986; von Elert & Jüttner, 1996, 1997). Fischerellin A, produced by Fischerella muscicola, is another compound acting against the PSII (Gross, Wolk & Jüttner, 1991) but here four different targeted sites have been identified (Srivastava, Jüttner & Strasser, 1998). In addition to its action against photosynthesis, fischerellin A is toxic for fungi at higher concentrations, although the mode of that action is still unknown (Hagmann & Jüttner, 1996).

Enzyme inhibition.  Many aquatic organisms produce extracellular enzymes that are essential for nutrition. Jüttner & Wu (2000) reported that 20% of the cyanobacteria isolated from freshwater biofilms in Taiwan could inhibit α-glucosidase activity. This may be a means to inhibit the hydrolysis of the mucilage produced by the cyanobacteria. The range of enzymes targeted by this activity and the implied compounds were not identified.

Cellular paralysis.  The cyanobacterium Anabaena flos-aquae can induce paralysis and thus faster settling of the cells of the competing motile green alga Chlamydomonas reinhardtii (Kearns & Hunter, 2001). This may create a competitor-free zone for the cyanobacterium.

Inhibition of nucleic acid synthesis.  Two alkaloids isolated from Fischerella sp. (12-epi-hapalindole E) and Calothrix sp. (calothrixine A) exhibit an inhibitory activity directed against the RNA polymerase of bacteria, fungi and green algae (Doan et al., 2000). This activity is strongly dependent on polymerase concentration and leads to growth inhibition because of protein synthesis inhibition. Calothrixine A also inhibits DNA synthesis.

ROS generation.  The violet pigment nostocine A, produced by Nostoc spongiaeforme, is highly cytotoxic for several micro-algae (Hirata et al., 2003). It has been found to accelerate the formation of reactive oxygen species (ROS) in the green alga C. reinhardtii. Inside the target cell, nostocine A is reduced specifically by intracellular reductants such as NAD(P)H. When the level of O2 is sufficiently higher than that of nostocine A, the reduced product of nostocine A is oxidised by O2 which generates the production of superoxide radical anion (Oinline image). Oinline image and the ROS subsequently derived from Oinline image may cause the cytotoxicity of the nostocine A (Hirata et al., 2004). An unidentified compound from cyanobacterium Microcystis sp. also induces oxidative stress in the dinoflagellate Peridinium gatunense. It inhibits carbonic anhydrase activity which simulates CO2-limiting conditions (Sukenik et al., 2002). In presence of light, this may lead to the formation of ROS because photosynthetic electrons could not be used to fix CO2. Depending on their concentration, these ROS and especially H2O2 may induce programmed cell death, a process close to apoptosis of animal and plant cells (Vardi et al., 1999). The alternative to death of the dinoflagellate cells is cyst formation.

Toxins

Most toxins are classified as hepatotoxins, neurotoxins or dermatotoxins after the symptoms they produce. However, because those symptoms have been mainly described in vertebrates, in the context of this review it is more relevant to classify them according to their chemical structures [cyclic peptides, alkaloids, lipopolysaccharides and polyunsaturated fatty acids (PUFAs) and their derivatives]. Some recent studies have focused on the effects of toxins on plankton and on macrophytes (Table 2). The effects of the toxin may be direct or indirect, linked to the metabolism of the molecule by the detoxification system.

Cyclic peptides.  Two toxins are cyclic peptides: microcystins and nodularins, microcystins being the most widely distributed toxins. Microcystins are produced by planktonic cyanobacteria belonging to the genera Anabaena, Microcystis, Planktothrix and by some species of the benthic Oscillatoria (Wiegand & Pflugmacher, 2005); nodularins are produced by Nodularia spumigena (Briand et al., 2003). These peptides contain unusual amino acids and show a strong structural variability: more than 75 structural variants of microcystin have been described to date. Microcystins and nodularins have been shown to be inhibitors of the serine/threonine protein phosphatases types 1 and 2A (MacKintosh et al., 1990; Honkanan et al., 1994). This activity has been demonstrated for mammals and higher plant protein phosphatases. The toxin–enzyme interactions are very strong, and binding is essentially stoichiometric. The concentration required to inhibit protein phosphatases in vitro is lower for nodularin than for microcystins (Ohta et al., 1994). Inhibition of protein phosphatases leads to hyperphosphorylation of proteins associated with the cytoskeleton and consequent redistribution of these proteins. In mammals and birds, the toxic effects of microcystins are almost restricted to the liver.

Besides these well-studied effects of microcystins against vertebrates, several cases of negative effects of these hepatotoxins against micro-algae or aquatic plants have been reported. The harmful effect observed in these cases may not be linked to the inhibition of protein phosphatases as for mammals but to the elevated formation of ROS which seems to occur in each case. Indeed, in the picocyanobacterium Synechococcus elongatus, the toxicity of the microcystin-RR seems to be linked to the induction of oxidative stress manifested by elevated ROS levels and malondialdehyde content (Hu et al., 2005). The oxidative stress induced in the dinoflagellate P. gatunense by microcystin-LR is linked to the activation of mitogen-activated protein kinases, enzymes known to play a role in cellular responses to biotic and abiotic signals in mammals and higher plants cells (Vardi et al., 1999; Vardi et al., 2002). Concerning aquatic macrophytes, several studies on the effects of microcystin-LR on Ceratophyllum dermesum demonstrated the existence of uptake of the toxin by the plant accompanied by a subsequent increase in ROS concentration and followed by an increase in the gluthatione S-transferase and several antioxidant enzyme activities, a growth inhibition and changes in pigments pattern (Pflugmacher, Codd & Steinberg, 1999; Pflugmacher, 2002, 2004). Interestingly, an increase in the amount of ROS was observed in rat liver after administration of cyanobacterial crude extract (Ding et al., 2000). The harmful effects of microcystins on mammals could be linked to both the inhibition of protein phophatases and the formation of ROS.

Alkaloids.  The alkaloid toxins include anatoxin-a (and homoanatoxin-a), anatoxin-a(s), cylindrospermopsins and saxitoxins. Anatoxins are mainly produced by Anabaena species, but also by Microcystis and Oscillatoria (Park et al., 1993; Sivonen & Jones, 1999). Cylindrospermopsins are produced by Aphanizomenon ovalisporum, Cylindrospermopsis raciborskii, Raphidiopsis curvata and Umezakia natans (Briand et al., 2003). Saxitoxins were first described in marine dinoflagellates but they have been recently identified in five freshwater cyanobacterial species: Aphanizomenon flos-aquae, Anabaena circinalis, C. raciborskii, Lyngbia wollei and Planktothrix sp. (Briand et al., 2003). Anatoxin-a and anatoxin-a(s), two unrelated compounds, both inhibit transmission at the neuromuscular junction. Anatoxin-a is a cholinergic agonist that binds to nicotinic acetylcholine receptor while anatoxin-a(s) is an acetylcholinesterase inhibitor with a mechanism similar to that of organo-phosphorus insecticides (Carmichael, 1994). Toxic effects observed on the aquatic plant Lemna minor may be linked to the metabolism of the toxin by the plant that may produce either reactive species of oxygen or a new compound toxic for the plant (Mitrovic et al., 2004). The toxicity of the hepatotoxin cylindrospermopsin (CYN) seems to be exerted through interference with protein/enzyme synthesis (Griffiths & Saker, 2003). The nucleotidic structure of CYN suggests that this toxin may have effects on DNA and RNA, and a some covalent interactions between CYN and DNA have been reported in treated mice, with significant DNA strand breakage (Shen et al., 2002). Saxitoxins bind to site 1 of the sodium channels in cell membranes, which blocks nervous transmission (Carmichael, 1994). To date there is no known effect of the saxitoxins on aquatic plants or on micro-algae.

Polyunsaturated fatty acids and their derivatives.  Compounds containing the αβγδ-unsaturated aldehyde structure (2,4-heptadienal, 2,4-octadienal from diatoms) (Wendel & Jüttner, 1996) act against herbivores. Cell division is blocked by the aldehydes, certainly because of microtubule de-polymerisation whereby tubulin cannot organise into filaments (Buttino et al., 1999). Moreover, no DNA replication can occur in presence of the aldehyde (Hansen, Even & Geneviere, 2004). In copepods, the polyunsaturated aldehyde induces a caspase-independent programmed cell death as revealed by cytochemical and biochemical approaches (Romano et al., 2003). In sea urchin, the aldehyde (2E,4E-decadienal) induces apoptosis and activates a caspase 3-like protease.

Regulation and influence of environmental and physiological factors

Allelopathic compounds

It has been shown for terrestrial plants that stress conditions decrease the importance of competition in favour of allelopathy in community structuring (Inderjit & Del Moral, 1997). An environmental stress (nutrients, light, temperature) may increase either allelopathic compound production or target sensitivity (Reigosa et al., 1999). The same phenomenon may exist for aquatic photoautotrophs: the importance of allelopathy is enhanced when the environmental conditions are suboptimal. In some cases this may be directly linked to the mode of action of the compound. The allelopathic compound produced by the cyanobacterium T. doliolum, which inhibits photosynthesis, is more toxic to Anabaena variabilis under light limitation (von Elert & Jüttner, 1996). This result points to the fact that experiments conducted under light saturation may underestimate the impact of allelopathic interactions given that many allelopathic compounds act against photosynthesis.

Little is known about the mechanisms regulating production or excretion. It has been shown for T. doliolum that regulation of the release of allelopathic compounds was decoupled from the release of dissolved organic carbon (DOC) (von Elert & Jüttner, 1997). Indeed, an increase in irradiance under P-limited condition led to elevated of DOC with no increase in the release of the allelopathic compound. This suggests the existence of two different regulation pathways, with allelopathic compounds having their own regulation pathways.

The influence of several parameters on allelopathic interactions has been studied in different cases. It appears that allelopathic interactions are affected in various ways by a great number of environmental factors. Nutrient concentration has an important influence on the interactions. Nutrient stress can enhance the production of allelopathic compounds by various algae (Ray & Bagchi, 2001; Rengefors & Legrand, 2001) and subsequently modify the equilibrium between taxa. For instance, under phosphorus-limited conditions, the release of allelopathic compounds by T. doliolum increases 30-fold (von Elert & Jüttner, 1996). Conversely, the allelopathic interaction between the dinoflagellate P. gatunense and the cyanobacterium Microcystis sp. seems to be independent of nutrient availability (Vardi et al., 1999). Another factor that must be taken into account is temperature. The antibiotic production by two cyanobacteria, Oscillatoria angustissima and Calothrix parietina is not proportional to the biomass but depends essentially on the temperature of the culture (Issa, 1999). A last factor in this non-exhaustive list is growth medium pH: the algicidal activity of Oscillatoria laetevirens was negatively correlated with pH (Ray & Bagchi, 2001).

The integration of the influence of environmental factors on the production of allelopathic compounds is essential for understanding the ecological role of allelopathy. It is necessary to know what compound production is in situ. Few studies focus on the variation in target sensitivity, even though it is the second part of the allelopathic interaction. Besides these abiotic factors, the intensity of the interaction may depend on biotic factors such as the concentration of the target or the composition of the bacterial community. Effects of the target on production of allelopathic compounds have rarely been studied, presumably because production of allelopathic compounds is often observed in the absence of the target. However, the production of antifungal molecules by the cyanobacterium Scytonema ocellatum was shown to be induced by fungal cell-wall polysaccharides (Patterson & Bolis, 1997). As allelopathic compounds may be metabolised by micro-organisms, their actual concentration may depend on the microbial activity. Finally, the cellular phase is also an important physiological factor that influences the interaction. The toxicity of Peridinium aciculiferum against competitors is maximal during stationary phase (Rengefors & Legrand, 2001). In contrast, several studies have indicated that the donor alga is more (or only) allelopathic when the culture is in exponential growth phase (Suikkanen, Fistarol & Granéli, 2004).

Toxins

Environmental parameters may influence both the production of the toxin (intracellular amount) and its release into the environment (extracellular amount compared with cellular concentration). Generally, toxins are released into the environment by cell lysis. Nevertheless, Rapala et al. (1997) showed in microcosm experiments that even if time is the most important factor that controls the release of microcystins into a growth medium, the concentration of dissolved toxins was increased by light flux above 25 μmol m−2 s−1 and by addition of nitrogen.

The effects of environmental conditions on toxin production have been reviewed by Sivonen & Jones (1999). In many cases, the production of freshwater toxin is negatively correlated with nitrogen concentration and positively with phosphorus concentration (Rapala et al., 1997; Kaebernick & Neilan, 2001). Highest production of toxin is generally found in optimal conditions for cell growth (Kaebernick & Neilan, 2001). However, the production of cylindrospermopsin by C. raciborskii is negatively correlated with growth rate (Griffiths & Saker, 2003), as for several allelopathic compounds. The identification of the genes of the microcystin synthetase, implied in microcystins synthesis, provided a new powerful tool for the investigation of the regulation of toxin synthesis. A recent study demonstrated that the transcription of two of these genes was controlled by light quality and initiated at certain threshold intensities (Kaebernick et al., 2000) which confirms the results of Rapala et al. (1997). These authors found no correlation between the transcription of one of these genes and cellular toxin content. This made them hypothesise that microcystins may be released from the cell and play a putative role under high light conditions.

Biotic factors, such as the presence of a competitor, may also influence the production of toxin (Vardi et al., 2002). The regulation pathways may depend on the toxin involved. For example, the extracellular production of anatoxin-a by Anabaena flos-aquae is increased while the production of microcystin is totally inhibited by high concentration of C. reinhardtii extracellular products (Kearns & Hunter, 2000).

Ecological roles

Allelopathic compounds

How can some species dominate the whole algal community? What factors control the dynamic of the benthic and planktonic algal communities and the formation and disappearance of blooms? The persistence of a species depends on its competitive capacity and species succession has often been explained as a consequence of competition. A species that produces allelopathic compounds will have an advantage over its competitors (Wolfe, 2000). Thus, allelopathy, as competition, should partly explain species succession. However, the problem of distinguishing between competition and allelopathy makes it difficult to evaluate the real importance of allelopathy in natural environments. Allelopathic effects can only be separated from those of competition in microcosm experiments.

Several cases have been well described where algal succession and the formation of blooms are related to the production of allelopathic compounds (Kearns & Hunter, 2001; Rengefors & Legrand, 2001; Vardi et al., 2002). A few examples are given here. Keating (1977, Keating 1978) combined field observations and laboratory studies to show that allelopathic interactions may be implicated in the establishment of bloom sequences in a eutrophic lake. Cyanobacteria that were dominant could inhibit both their predecessors and their successors and there was a negative correlation between diatom blooms and cyanobacterial blooms. More recently, similar results were found for diatom-cyanobacteria succession in a eutrophic lake in Japan (Takamo et al., 2003). By comparing the growth of the cyanobacterium Phormidium tenue in the presence of diatoms with and without germanium, a specific growth inhibitor of diatoms, the authors demonstrated that cyanobacterial development was restrained by the production of inhibitory compounds. In the lake, the decrease in diatoms due to consumption of the available phosphorus allowed the development of the cyanobacteria previously inhibited. A freshwater bloom-forming green alga, Botryococcus braunii, excretes free fatty acids which have adverse effects on various phytoplankton and zooplankton species (Chiang, Huang & Wu, 2004). The presence of these species during B. braunii blooms was negatively correlated with their sensitivity to the fatty acids. The end of the bloom coincided with a decrease in the production of free fatty acids. These results suggest a relationship between the formation and disappearance of the bloom and the production of allelopathic compounds that eliminate both competitors and predators.

Allelopathy may also be a way to compensate competitive disadvantage (low growth rate, low nutrient uptake). This is the case for the freshwater dinoflagellate P. aciculiferum which produces allelochemicals that would compensate for the disadvantage of its large size in terms of nutrient uptake and help it to dominate in winter the phytoplankton biomass (Rengefors & Legrand, 2001). Moreover, the lysed target cells release nutrients that can support the dinoflagellate growth.

Toxins

The ecological role of toxins is still debated. Several hypotheses have been proposed, in particular for the microcystins, but to date no consensus has emerged. The term ‘toxin’ groups some molecules of quite different chemical nature and with various biosynthetic pathways, and thus they may have diverse functions. From an evolutionary viewpoint, and given the high costs of their production supported by the cell, toxins must be presumed to have an ecological role. Most research in this field has focused on microcystins and several hypotheses have been proposed to explain their production. Initially it had been proposed that the toxins may have originally had a critical function that is now lost (Carmichael, 1994). This is supported by the fact that the activity of microcystins is directed against the protein phosphatases that regulate eukaryote proliferation, and by the absence of participation of the toxins in cell function and cell division. Nevertheless, given the important cost of production linked to the enzymatic complexes involved in their synthesis, it seems probable that toxins have an actual function in cyanobacterial physiology or ecology. A second early hypotheses was that toxins may be predator defences, as toxic or deterrent compounds (DeMott, Zhang & Carmichael, 1991). Indeed some studies demonstrated that microcystins can be toxic for zooplankton (copepods, cladocerans) or may induce avoidance behaviour (Kurmayer & Jüttner, 1999; Ghadouani et al., 2004). This is also true for the toxin cylindrospermopsin (Nogueira et al., 2004). A third hypothesis is a role for the toxin in the regulation of light harvesting and chromatic adaptation. This is supported by the genetic study of Kaebernick et al. (2000) who demonstrated regulation by light quality and intensity of the genes involved in microcystin synthesis. A fourth hypothesis is that microcystins may be iron-scavenging molecules and thus may be associated with iron transport (Utkilen & Gjolme, 1995). A Fifth hypothesis is that toxins represent some storage substances (Carmichael, 1997). Toxins have also been proposed to act as allelopathic compounds (Christoffersen, 1996; Pflugmacher, 2002). A final hypothesis was proposed by Sedmak & Kosi (1998) who suggested that microcystins could act as growth regulators helping cyanobacteria to multiply and giving them a better opportunity for successful adaptation. Their hypothesis is based on experiments that showed stimulating effects of microcystins on the growth of various green algae and cyanobacteria.

Among the hypotheses proposed to explain the production of microcystins, some may be equally applied to the others toxins: protection against predators and allelopathic function.

In addition to the issue of understanding the role of toxins, a big question is to explain the co-existence of toxic and non-toxic strains. Non-toxic strains may either take advantage of the production of toxins by co-occurring strains, without supporting the costs of synthesis, or they may produce compounds with an activity similar to that of toxins but that are not toxic to the animals that are the focus of most toxicity testing. Recently, peptide production by 18 clonal strains of Planktothrix sp. from a single water sample has been investigated by MALDI-TOF mass spectrometry and HPLC (Welker et al., 2004). Each strain appeared to produce between three and eight major compounds comprising microcystins and other known peptides. The putative biological activity of most of these peptides remains unknown but could be the same for both toxic and non-toxic compounds. Thus, microcin SF608, a peptide produced by Microcystis sp., induced typical stress reaction in aquatic plant and zooplankton, as microcystin-LR (Wiegand et al., 2002).

The production of toxins may be a protection mean against predators. Some PUFAs and their derivatives have been shown to have anti-predator properties (see review by Watson, 2003). When the freshwater anostracan Thamnocephalus platyurus fed on diatom-dominated biofilms it showed high mortality linked to the production of various PUFAs with 5,8,11,14,17-eicosapentaenoic acid being responsible of most of the toxicity (Jüttner, 2001). Such fatty acids are only produced by the hydrolysis of lipids in the environment or within the digestive tract of the grazer when cell walls are disrupted. This makes the use of PUFAs a very efficient strategy for grazer defence: basic cellular components (e.g. membrane lipids) are rapidly transformed into highly toxic molecules when the alga is grazed. This means no additional cost for the synthesis of a specialised defence molecule or no problem of autotoxicity and of transfer to the target. Within the PUFAs derivatives produced by diatoms, αβγδ-unsaturated aldehydes (2E,4Z-decadienal, 2,4-heptadienal) have adverse effects on grazers, mainly on the next generation (Miralto et al., 1999; Pohnert, 2000).

Toxins or allelopathic compounds?

Because of their potential health hazard, most toxic compounds have been described and studied as toxins. Nevertheless, as a function that could be fulfilled by both toxins and other related compounds, allelopathy has received much attention in the recent years. Defence against grazers and inhibition of competitors can confer strong competitive advantages to the producer which may have been sufficient for the selection of toxin-producing strains. It should be noted that nostocyclamide, a cyclic peptide having a biosynthesis pathway close to that of microcystins and nodularins (Kaebernick & Neilan, 2001), has been identified and studied on the base of its allelopathic properties (Todorova & Jüttner, 1995).

Several cases where toxins act as allelopathic compounds have been reported. The negative effect of Anabaena flos-aquae on the green alga C. reinhardtii is mediated by both microcystin-LR and anatoxin-a (Kearns & Hunter, 2000). Singh et al. (2001) reported that purified microcystin-LR from Microcystis aeruginosa had a negative effect on the growth of several green algae and cyanobacteria. Growth of S. elongatus was inhibited by microcystin-RR (100 μg L−1) with a decrease in photosystem II efficiency (Hu, Liu & Li, 2004). The effects of microcystin-LR against the aquatic plant Ceratophyllum demersum have been fully described by Pflugmacher (2002, 2004). He demonstrated that the toxin could inhibit growth and had adverse effects on photosynthesis and pigment pattern at environmentally relevant concentrations (5 μg L−1). The plant exhibited an oxidative stress while detoxication and antioxidative enzymes were induced (Pflugmacher, 2004). Pflugmacher pointed to the fact that gluthatione S-transferase can recognise microcystins as natural substrates and thus initiate the detoxication/elimination process. This emphasises the fact that an allelopathic role of toxins could be considered in some cases. Microcystin-LR was also found to influence growth and morphology of the aquatic plant Spirodela oligorrhiza (Romanowska-Duda & Tarczynska, 2002). The inhibitory effects were observed with cellular extracts containing 0.334 mg L−1 of microcystin-LR and with 0.1 g mL−1 of commercial-grade microcystins-LR. In some of these studies, the concentrations of microcystin were rather high compared to those measured in the field (below 10 μg L−1) (Sivonen & Jones, 1999). Nevertheless the possibility that elevated concentrations occur in the micro-environment surrounding algal cells must be considered. Moreover, cellular extracts containing toxins are often more active than purified toxin, which suggests that cellular extracts contain a mix of active the toxin that may act synergistically.

The effects of anatoxins on aquatic plants have been less studied. Negative effects on an aquatic plant have been reported (Mitrovic et al., 2004). A 4-day exposure of the free-floating plant L. minor to anatoxin-a led to an increase in both peroxidase activity and gluthatione S-transferase activity (two detoxification enzyme activities) while photosynthetic oxygen production was reduced. In that study the mode of action of the toxin is unknown. Anatoxin-a concentrations required for the observation of a significant effect were quite high (25 μg mL−1) compared to natural concentration (rarely above 3 μg mL−1), which limits the relevance of the described interaction.

In this review we have defined predator defences as allelopathy. Under that definition, there are many examples that support the allelopathic role of certain toxins. Cylindrospermopsis raciborskii appeared to reduce the fitness and the growth of juvenile D. magna, which was partly due to the production of cylindrospermopsin; a non toxic C. raciborskii strain affected the same crustacean zooplankton species to a lesser extent (Nogueira et al., 2004). In vitro experiments demonstrated that microcystin-LR can induce oxidative stress enzymes and cause death to D. magna (Wiegand et al., 2002). As for activity against photoautotrophic organisms, toxins may also act synergistically with other compounds. Indeed, the negative effect of PUFAs on D. magna was enhanced by the addition of microcystin-LR at a concentration at which it was not active alone (Reinikainen et al., 2001). The toxicity of microcystin-LR may also be more pronounced when it is delivered via food rather than via water (Reinikainen, Ketola & Walls, 1994).

Effects of toxins on predators and competitors very much depend on the target tested; results cannot be generalised. In many studies a high concentration of the toxin is needed to observe an effect, one reason why some authors exclude the hypothesis of an allelopathic role of toxin. However, as stated before, toxins may act synergistically with other compounds, at much lower concentrations. Some studies also implicate toxins in algae–algae interactions (Sivonen & Jones, 1999; Kearns & Hunter, 2001; Vardi et al., 2002). Moreover, in some cases, toxin production is affected by the presence of various photoautotrophic organisms which is consistent with a role for these compounds in biotic interactions (Kearns & Hunter, 2000; LeBlanc, Pock & Aranda-Rodriguez, 2005). Finally, many studies have demonstrated that toxins can have adverse effects on grazers (see review by Wiegand & Pflugmacher, 2005). Among the other secondary metabolites produced by cyanobacteria, some share the same biosynthesis pathway as toxins. Those compounds are not toxic for vertebrates and thus they have not been studied as toxins. Nevertheless they may have a similar function in predator defence or competitor inhibition. All these results support the view that toxins may be allelopathic compounds. Although further research is needed to clarify the allelopathic effects of toxins, the allelopathic hypothesis remains relevant.

Ancillary