This subject has been recently reviewed by Knoll and collaborators [17]. The evolutionary history of phytoplankton has been studied through both morphological fossils (well-preserved structures such as cell walls, scales or cysts available for some taxa) and molecular biomarkers such as lipids or nucleic acids. The data suggest that prokaryotes have governed ecosystems during the Proteozoic. Primitive cyanobacteria or other prokaryotic organisms such as anoxygenic photoautotrophs (that might have used H₂ as final electron acceptors) could have dominated the marine flora in the Archean. Biomarkers found in shales (which biosynthesis is known to require oxygen) and observed in extant cyanobacteria suggest that the later had evolved oxygenic photosynthesis by 2.7 Gy, before the initial accumulation of free oxygen in the atmosphere 2.3–2.4 Gy ago. Limited data from proterozoic rocks suggest that cyanobacteria and other photosynthetic bacteria dominated primary production at that time while biochemical evidences suggest that green algae began to play a role as major primary producers during the late Proterozoic and Cambrian. This primitive phytoplankton assemblage prevailed during the Mesozoic, until the radiation of the chlorophyll a+c algae namely the diatoms, dinoflagellates and haptophytes. Fossils clearly suggest that these groups became dominant during the Mesozoic. The cyanobacteria and green algae, which appeared then to be less important in coastal plankton, were still probably co-dominant or dominant in the oligotrophic central gyres and dominate the modern picophytoplankton size class.

The mechanisms by which photosynthesis was acquired, and oxygenic photosynthesis was established in primitive cyanobacteria is still poorly understood [18]. However, the process of photosynthesis acquired by cyanobacteria, spread across the eukaryotic tree of life through the process of endosymbiosis. Early in the evolution of eukaryotes, a heterotrophic eukaryote engulfed and retained a cyanobacterium, converting it into a plastid after intracellular gene transfers between the primitive host and its symbiont. This event, the “primary endosymbiosis” that probably occurred a single time [19,20], gave rise to the extant Archaeplastida which include the Glaucophyta, Rhodophyta (red algae) and Chloroplastida (green algae and land plants). The primary plastid was suggested to be established around 1.25–1.60 Gy ago, during the Proterozoic [4]. Members of these lineages were themselves engulfed by non-photosynthetic protists probably in separate endosymbioses and converted into plastids [20]. A secondary endosymbiosis event (0.8–1.2 Gy, [21,22]) with a primitive red algae gave rise to the cryptophytes, haptophytes, photosynthetic stramenopiles (or heterokontophytes) and peridinin-containing dinoflagellates. Two separate endosymbioses during which the plastids of green algae were integrated are at the origin of euglenophytes and chlorarachniophytes [4]. Some of the extant dinoflagellates arose from a tertiary endosymbiosis in which an alga containing a plastid originating from a secondary endosymbiosis was retained by a heterotrophic protist and reduced to a plastid. In modern oceans, photosynthetic eukaryotes that resulted from a secondary endosymbiosis with a red algae (the so-called “red” lineage), were extremely successful from the Mezozoic era to the present days [23]. According to Falkowski and collaborators [23] this success could be explained by the better transportability of red algal plastids. Indeed, the larger amount of key protein-coding genes in these plastids compared to the green algal plastids [24], may have increased the transfer probability through secondary endosymbiosis, in new, phylogenetically diverse hosts. Theses hypotheses however, have been questioned [25] and are in direct conflict with the “chromalveolate hypothesis” which proposes that all algae believed to possess secondary red plastids acquired them by a single common endosymbiosis. Overall, endosymbiosis had a major influence on the early evolution of algae. Indeed, this process was not only at the origin of lateral transfer of photosynthetic genes, it also enriched nuclear genomes with cyanobacterial genes (18% of nuclear encoded genes in Arabidopsis are estimated to have been acquired from the primitive cyanobacterial endosymbiont [26]).

Diatoms can be planktonic, benthic, epiphytic, epizoic, endozoic, endophytic and can also live in air [27]. Approximately 40% of all marine phytoplankton described species are diatoms and they are of crucial importance from an ecological and biogeochemical point of view, especially in nutrient-rich systems [23]. Diatom cells are encased within a special cell wall made of silica, the frustule, which is well preserved in the fossil record. They often form chains and colonies. They are traditionally divided into 2 groups. The centric diatoms have valve striae (rows of pores) arranged basically with central symmetry. In this case, the symmetry can be unipolar (radial centrics) or multipolar. Pennate diatoms have valve striae arranged basically in relation to a line. Within the latter, the frustule of raphid pennates has a slit called a raphe. Different phylogenetic studies showed that “radial centrics” represent the deepest branch while “bi- and multi-polar centrics” appeared later. Both groups form clades or sometimes ill-supported basal ramifications, but many well-supported terminal lineages [28]. All pennates, emerged from bipolar centrics, and form a well-supported clade. This group differentiated into the “araphids” and the subsequent “raphids”, rendering the araphid pennates paraphyletic. Molecular phylogenies of diatoms are in general agreement with evolutionary hypotheses proposed based on fossil records [29,30].

Species belonging to the Dinophyta can be planktonic, benthic, symbionts or parasites [31], but the majority of photosynthetic species are free-living and planktonic. The main characteristic of dinoflagellates is the presence of a large nucleus, with permanently condensed chromosomes. A typical cell displays 2 flagella, one in an equatorial groove, the cingulum, and the other projecting toward the posterior of the cell, often in a groove, the sulcus (dinokont species). Some species (desmokonts) have both flagella at the cell apex. Dinoflagellates can harbour thick cellulose plates located in alveoli under the plasmalema and forming a rigid theca (armoured dinoflagellates). These plates can also be inconspicuous or absent in the unarmoured species. Less than half of all dinoflagellates are photosynthetic [32,33], the majority of them harbor peridinin as the main accessory pigment [34]. All the photosynthetic dinophytes belong to the class Dinophyceae sensu stricto [35]. Dinoflagellates phylogeny is not consensual [36]. Phylogenies inferred from the ribosomal genes often present problems of long branch attractions and are sensitive to analytical method [37,38]. Moreover, dinoflagellates rDNAs present regions with different rates of evolution, containing stem regions with sites that do not evolve independently [39]. Phylogenies based on sequences of plastid genes are often inappropriate, due to the complex endosymbiotic history of these organisms [40], and because the plastid genome is highly reduced, organized in minicircles, and exhibits a strikingly high rate of evolution and a tendency to unique rearrangements [41,42]. However, the Dinophyceae sensu stricto often form a separate clade within the Dinophyta. This clade includes the Dinophyceae with a peridinin-containing plastid, that is thought to be the ancestral plastid of Dinophyceae, but also some non-photosynthetic species that presumably lost their plastids, as well as photosynthetic species without peridinin that acquired other plastids through a secondary or tertiary endosymbiosis [36]. Dinokont/desmokont or armoured/unarmored species do not form separate clades. Orders and genera within the phytoplanktonic dinoflagellates are not always supported by molecular phylogenies. For example, the order Gymnodiniales (dinokont, unarmoured cells) is polyphyletic according to Murray and collaborators [39] while the analysis of LSU rDNA sequences resulted in the splitting of the genus Gymnodinium into four genera [37]. Evolutionary hypotheses that emerge from molecular phylogenies are not easily compared to fossil records. This is mainly due to the poor fossilization of vegetative cell walls. Resistant cysts (resting stage) are found in fossils records, but only 13% of modern dinoflagellates are known to form them [43]. Despite many phylogenetic studies, the evolution of the major dinoflagellate lineages remains uncertain and morphogenetic studies are necessary, especially concerning the small unarmoured species.

Haptophytes are mainly composed of small planktonic unicellular species, occurring single or in colonies. Heterotrophy is not common. Cells usually have 2 flagella, with an associated structure, the haptonema sometimes used for the cell anchoring or prey capture [44]. Organic scales often cover the cell body surface. The division Haptophyta is composed of two classes: the Pavlovophyceae (with 2 unequal flagella) and the Prymnesiophyceae (with 2 more or less equal flagella). Within the Prymnesiophyceae, the Calcihaptophycidae are potentially calcifying haptophytes [45]. This group includes the coccolithophores, which are covered by small regular calcareous plates (coccoliths) which are important in biogeochemical cycles since they are responsible for about half of all modern precipitation of CaCO₃ in the ocean [46]. Haptophytes belong to one of the deepest branching groups in the phylogeny of the eukaryotes [47]. Comparative analyses of rDNA sequences also showed that the Calcihaptophycidae are monophyletic, but whether coccolithogenesis has single or multiple origins is still discussed [45]. The lack of geological records makes reconstruction of the primitive evolutionary history of uncalcified haptophytes very difficult.

Within the super-group Archaeplastida, the marine phytoplanktonic green algae have representatives in the classes Trebouxiophyceae and Prasinophyceae (Fig. 1) that are distinguished based on morphological and ultrastructural characters (flagellar types and number, flagellar insertion structures, cell wall, type of mitosis, [48]), as well as molecular phylogenies [49]. The ubiquity and contribution of green algae to marine phytoplanktonic assemblages and more precisely to the picoplanktonic fraction has been established using microscopy [50], pigment analyses, and molecular probing [8,16]. The prasinophytes, which is the most diversified group among marine phytoplankton green algae, occupies a critical position at the base of the green algal tree of life. Members of this group are supposed to have retained primitive characters such as the presence of organic scales on the cell bodies. The diversity in their cell shapes, flagellar number, flagellar apparatus organization, scale morphology and cell division features led some authors to suggest that they were not a monophyletic group (Lewis et al. [48]). Molecular phylogenies have echoed the morphological diversity and identified at least seven phylogenetic groups [49]. Genetic studies on the diversity of picoplanktonic green algae investigated by culture independent approaches in the Mediterranean Sea [16] confirmed the prevalence of the prasinophytes within the Archaeplastida (99% of the sequences detected), and revealed new clades, as well as an important diversity at the sub-genus level.

Both filamentous (heterocystous or not) and unicellular forms of this group have representatives in the oceans. Marine cyanobacteria have a global ecological significance, not only for the carbon but also for the nitrogen cycles. Marine phytoplanktonic genera are dispersed within the Cyanobacterial-tree. Two closely related genera, Prochlorococcus [51] and Synechococcus [52] are particularly abundant and ubiquitous in the ocean but do not fix atmospheric nitrogen. Each of these genera contains distinct phylogenetic clades representing physiologically and ecologically distinct populations (or ecotypes [53,54]). At least two marine, nitrogen-fixing genera are also known: the colony-forming, filamentous Trichodesmium and the coccoid Crocosphaera, both restricted to tropical waters [55]. The genetic variability within the latter genus was recently be found to be remarkably low [55]. More diversity, at different taxonomic levels, is probably to be discovered in the marine planktonic cyanobacteria, both for non-N₂-fixing and N₂-fixing taxa [56,57].

Dinoflagellates, diatoms and coccolithophores, each with plastids derived from red algae by secondary endosymbiosis rose to ecological prominence during the Mesozoic and continue to dominate the phytoplankton biomass. Hypotheses involving competition for light and nutrients as major driving forces in phytoplankton evolution have been proposed to explain this prominence [23,28]. The presence of specific accessory pigments [34] has been proposed as an important selective character; these would have allowed better light absorption, protecting the cells from light saturation, as well as allowing photoacclimation [40]. Moreover, these algae also possess efficient mechanisms for uptake and internal recycling of nutrients and can sometimes take up particulate or dissolved organic matter or be associated with diazotrophic organisms [28]. The ability to form resting stages has also been considered to provide an important selective edge [28]. Smetacek [58] argues that the enormous diversity of shapes and lineages must do more than improve the photosynthetic efficiency of chloroplasts, and that evolution in the plankton is ruled by protection against predation and pathogens (viruses and bacteria), and not only competition.

The reduction of the size, as well as the acquisition of efficient nutrient absorption pathways are probably key driving forces for the small phytoplankton size classes. These aspects of evolution are started to be elucidated as data from the fields of ecology and genomics are being integrated [59].

The accurate circumscription of species is an essential requirement for biodiversity assessments, as well as for a proper understanding of their ecology and evolutionary history [62]. Despite long running debates, there is still no consensus and various species concepts have been adopted, often implicitly. While the traditional morphological species concept dominates in the field of phytoplankton taxonomy, several authors have expressed a need for a clearer definition that would include genetic circumscription and mating delineation [62] because morphology is too influenced by external conditions and can change according to life stages. Mating delineation that would allow the definition of biological species as “groups of interbreeding natural populations that are reproductively isolated from other such groups” [63] is especially problematic for phytoplankton organisms, as (1) distinctive morphological features are not always easy to identify (Fig. 4) and (2) sexual reproduction is completely unknown in several phytoplankton groups.

During the last decades, the genetic diversity within morphologically-defined phytoplankton taxa has been explored and has revealed a wide cryptic diversity. Coccoid picoplankters are extreme cases in this respect. Their minute size and simple spherical morphology limit the taxonomic level to which these cells can be identified to phyla or classes [64]. Flagellates that do not harbour hard structures are also extremely difficult to identify to the species level. Beyond these extreme cases recent genetic studies have also pointed out to the cryptic diversity within well-characterized species. Indeed, clear genetic clades are commonly found within widespread phytoplankton species. However, the ecological and evolutionary significance of these genetic variations found within well-known morphological taxa and in culture independent molecular survey [16], is often not well understood.

In some cases, the careful examination of the morphology of the clades identified genetically have revealed the existence of pseudo-cryptic species, i.e. species that can be differentiated only on very subtle morphological features. For example, clear genetic clades corresponding to fine scale morphovariants suspected to represent eco-phenotypes or ecotypes, were identified using the 18S rDNA and tufA genes within the coccolithophore Calcidiscus leptoporus [65]. Similarly, detailed analyses of ITS sequences and morphological features allowed to distinguish 3 clear genetic groups within the diatom Pseudo-Nitzschia delicatissima [66].

Genetic variability associated with specific biogeographical origin, can also be found in cosmopolitan morphospecies of diatoms such as Skeletonema costatum [67], Skeletonema marinoi [68] or dinoflagellates such as Scrippsiella spp. [69].

In some other cases, if the genetic clades identified are indistinguishable based on detailed morphological examination and appear to be cosmopolitan, fine scale analyses of their distribution shows that they occupy specific niches in the marine ecosystems. This is the case for Micromonas pusilla, a widespread green picoplankter. Between 3 and 5 clades were recently identified within this species, with genetic divergence higher than divergences estimated between traditional genera [70]. While those clades are commonly found in sympatry, they seem to occupy specific ecological niches and to compete in some environments [71]. Most interestingly in the context of the delineation of the phytoplankton ideal species, the mating compatibility of genetic variants was recently studied for the diatoms Pseudo-nitzschia delicatissima [62] and Sellaphora pupula [72]. These studies showed that mating compatibility occurred only for strains presenting highly homologous ITS2 sequences.

More studies combining genetic analyses on several genes and ecological, morphological, physiological and biogeographical analyses have to be conducted in order to clarify the taxonomic status of many phytoplankton species. These studies would also help formulating a new species concept that would in return allow: (1) much more satisfactory biodiversity assessment and monitoring; (2) advances in our knowledge of the ecology, physiology and genetics of phytoplankton; and (3) understanding the evolutionary processes that prevail in the ocean.

The ecological and evolutionary processes of speciation and the mechanisms by which diversity is maintained in the pelagic microbial realm are poorly understood. Indeed, the coexistence of dozens of phytoplankton species in natural assemblages where only a few resources are potentially limiting (the so-called paradox of the plankton, [73]) has puzzled the biologists for decades. The discoveries of a previously unsuspected diversity in the picoplankton size fraction, and of large genetic variations within widespread taxa amplifies this paradox. In phytoplankton, many populations appear to be so large and so dense that allopatric divergence and speciation seem unlikely, or would be too slow to be able to account for the presently observed diversity. Moreover, it is likely that these organisms do not possess complex behaviours associated with reproduction that would limit reproductive isolation. Recently, analyses of the fossil records for cosmopolitan taxa [74], genomic analyses [59] and theoretical modelling [75] have provided new evidences supporting the potential widespread occurrence of sympatric speciation in the pelagic environment. These findings are important for the understanding of speciation in the pelagic realm. In parallel genomic studies conducted recently on individual picophytoplanktonic strains have shown the occurrence of new gene acquisition through horizontal transfer both in eukaryotes [59] and cyanobacteria [76] and have provided insights into the mechanisms of species divergence and niche partitioning through the gain or loss of genes.