1. Intracellular transport
All cells must deal with the problem of intracellular transport. Unicellular phytoplankton have a minimum cell size of just 1 µm3, which is perhaps proximally influenced by ecological factors, but ultimately dictated by nonscalable aspects of cell structure (Raven, 1998a). Examples of nonscalable structures that could restrict further decreases in size to much below 1 µm3 are proteolipid bilayer membranes based on C16–C18 fatty acids (hence c. 8 nm thick), genomes which, for photolithotrophs, probably cannot have fewer than c. 1400 genes, and ribosomes. Individual proteins are also nonscalable but would not restrict decreases in cell size until much smaller hypothetical cell sizes. As cell size increases, the surface-to-volume quotient decreases and the average transport distance within the cell increases, so diffusion becomes increasingly inadequate as a means of maintaining constant solute concentrations throughout the cytosol of the cell (Nobel, 2005). The influence of cell shape on intracellulular transport is considered in the Supporting Information (Text S1).
The largest diatom cells, along with the largest desmid cells (charophyceans which are a few tens of µm thick and 0.4 mm in diameter), lack the capacity to distribute solutes using cytoplasmic streaming. Both groups have a single nucleus that is located near the centre of the cell. For diatoms, it is in the centre of the vacuole and connected to the peripheral cytoplasm by cytoplasmic strands. The movement of nuclear transcripts from a single source might create problems for the transport of mRNAs and proteins – particularly proteins with large molecular masses and hence low diffusivities – through the cytosol to appropriate sites such as plastids and mitochondria. Maintaining solute fluxes at the rates required for growth over distances of up to hundreds of µm requires larger gradients (expressed as moles per cubic metre concentration difference per metre of pathway) because average transport distance, which scales with cell volume, V, as ∝V1/3, is not entirely offset by declines in specific growth rate with increasing cell size (∝V−α, α < 1/3; see Section V below). Vacuolation helps to increase the surface area for interception of photosynthetically active radiation (PAR) and for nutrient influx per unit metabolizing volume (i.e. that part of the cell not occupied by the vacuole; Raven, 1997a), but does not explicitly help with the problems of intracellular transport from the nucleus to the extremes of the cell.
Larger transport distances may also increase the difficulty of synchronizing the expression of nuclear and organelle (chloroplast and mitochondrion) genes. The majority of endosymbiont genes that have been retained – mainly for organelle functions (Reyes-Prieto et al., 2006) – have been transferred to the nucleus. Only a small minority of genes are retained in the organelle, perhaps in part because rates of mutagenesis can be relatively high (Allen & Raven, 1996). Synchronizing of gene expression is made difficult because organelle genomes are only a few µm from the site at which the resulting genes function, while nuclear genes can be a few hundred µm from the site of function in organelles. For chloroplasts, this is true regardless of whether a cell has many small chloroplasts or a single, large, reticulate chloroplast, because multiple copies of the chloroplast DNA are dispersed throughout a single large plastid (Kuroinia et al., 1981).
Further quantitative work is needed to determine the extent to which these potential synchronization problems in large cells actually occur and, if they do, whether these are significant in the context of the specific growth rate of Ethmodiscus of ≤ 0.39 d−1 (Villareal et al., 1999). In this context, it is of interest that larger cells of colonial Hydrodictyon, which are planktonic algae with cell volumes (mainly vacuole) of up to several mm3, are multinucleated and lack cytoplasmic streaming (Van den Hoek et al., 1995; Graham & Wilcox, 2000). Larger, benthic differentiated unicellular (or perhaps better described as acellular) ulvophycean green algae, such as Bryopsis and Caulerpa (Caulerpales = Bryopsidales; Graham & Wilcox, 2000), are generally also multinucleate, but do possess cytoplasmic streaming (Raven, 2003). However, algae such as Hydrodictyon have a regular spatial arrangement of nuclei, each surrounded by a similar-sized cytoplasmic ‘domain’ that may be determined by the cytoskeleton, whereas the arrangement of nuclei within members of the Caulerpales is random and not associated with the cytoskeleton (McNaughton & Goff, 1990). Taken together, these findings suggest that larger cells may maintain greater numbers of nuclei as a means of reducing intracellular transport distances, but that cytoplasmic streaming may be important when the size of nuclear ‘domains’ is variable. By contrast, members of the Dasycladales (e.g. Acetabularia; Van den Hoek et al., 1995; Graham & Wilcox, 2000) are uninucleated through most of their life cycle despite being relatively differentiated and having a maximum dimension of tens of mm (see Serikawa et al., 2001). These organisms possess cytoplasmic streaming and it is thought that the cytoskeleton (actin filaments) is involved in transport and/or localization of mRNA (Vogel et al., 2002). These roles of cytoplasmic streaming are additional to their involvement in transferring low molecular mass solutes within large differentiated organisms with spatially localized functions and resource sources (Raven, 2003).
2. Diffusion boundary layers
A diffusion boundary layer exists around all objects in a fluid medium; the thickness of this layer is the distance from the surface within which all solute and solvent movement normal to the surface of the organism is by diffusion (Denny, 1992). Theoretical considerations (e.g. Denny, 1992; Raven, 1998a) predict that the external diffusion boundary layer thickness is equal to the radius of the organism for a spherical cell, although the boundary layer thickness is smaller than this prediction for radii > 50 µm. Whether this layer effectively leads to growth limitation depends on the balance of nutrient concentration in the bulk medium and the rates of diffusivity and nutrient transport at the cell surface. Diffusive limitation of growth when the concentrations of nutrients such as nitrate and phosphate are low is, all else being equal, more likely for colonies composed of cells of < 50 µm radius than for unicells of the same size as the cells in the colony. This is because, as we shall see in Section VI, the potential specific growth rate decreases with increasing cell or colony size less rapidly than would maintain the relative importance of diffusive limitation of nutrient supply by the diffusion boundary layer in the overall uptake process.
The greater importance of boundary layers is consistent with the generally observed increase in the concentration of nutrient needed to give half the maximum rate of nutrient uptake with increasing cell size for unicells (Eppley et al., 1969, Litchman et al., 2007; cf. Smith & Kalff, 1982). Even allowing for any decrease in maximum specific growth rate with increasing cell size (Section VI) it would be expected that the proportionally greater restriction on the rate of uptake by diffusion through boundary layers would have the effect of decreasing the apparent affinity of the entry of solutes that reach saturation as the concentration increases. Such saturation can arise from either a plasmalemma-located transporter (e.g. of CO2 or in photosynthesis by most algae) or diffusive entry through nonsaturating mechanisms (e.g. by lipid solution) followed by assimilation using an enzyme (e.g. in the minority of algae that lack inorganic carbon concentrating mechanisms (CCMs)) (Raven et al., 2005a,b; see also Raven et al., 2008 for a discussion of NH3/ entry and assimilation). There are, however, serious methodological problems in determining half-saturation constants for growth, with those measured experimentally being somewhere between the real (free of experimental problems) half-saturation constant for growth (Kg) and that for the actual transporter (Kt); Kg is likely to be very much less than Kt (Flynn, 1998). In the absence of systematic measurements of the affinity of transporters and enzymes in a more resolved system with minimal and constant restrictions imposed by diffusion of substrate, it is not possible to be sure whether the diffusion boundary layer effects account for all the apparent decrease in affinity in vivo with increasing cell size, or whether there is a relaxation in selection for a high affinity of the transporters or enzymes for their substrates when such affinities cannot be fully expressed in vivo because of diffusive limitation or because of regulation of the transporters (Flynn, 1998).
Direct measurements of boundary layer thickness in phytoplankton organisms using microsensors have only proved possible for relatively large colonial planktonic organisms such as Phaeocystis (Ploug et al., 1999a,b), Trichodesmium (Carpenter et al., 1990), a symbiotically photosynthetic freshwater ciliate (Sand-Jensen et al., 1997), unicells such as planktonic symbiotic Foraminifera (Jørgensen et al., 1985; Rink et al., 1998; Köhler-Rink & Kühl, 2005), and a large centric diatom (Kühn & Raven, 2008).
The diffusion boundary layer around Phaeocystis colonies was found to be consistent with theoretical expectations in the study of Ploug et al. (1999a,b). Differences in phosphate affinity between single filaments and ‘puff’ and ‘tuft’ colonies of filamentous Trichodesmium also appear consistent with theory, at least qualitatively (McCarthy & Carpenter, 1979; Fu et al., 2005). The low affinity (here measured as the reciprocal of the half-saturation concentration, but more appropriately expressed as the substrate-saturated rate divided by the half-saturation concentration) for inorganic carbon (not saturated in air-equilibrated seawater) in diazo-photolithotrophic growth of Trichodesmium might also be related to diffusive boundary layers around ‘puffs’ and ‘tufts’ (Hutchins et al., 2007; Levitan et al., 2007; Ramos et al., 2007). This is because this marine β-cyanobacterium has a subset of the CCM components typical of freshwater β-cyanobacteria (Badger & Price, 2003; Badger et al., 2006), which might be expected to give saturation of photosynthesis in air-equilibrated seawater if there were no significant boundary layer limitations. Tchernov & Lipschultz (2008) reported changes in stable carbon isotope discrimination (δ13C) in Trichodesmium that are consistent with enhanced diffusion limitation of CO2 supply in larger colonies.
Evidence for such boundary layer effects has, however, been mixed. For example, specific growth rates have been found to be higher for phytoplankton colonies than for unicells in the studies of Veldhuis et al. (2005: see Section VI for further work on Phaeocystis) and Wilson et al. (2006). It is noted that, in these studies, growth was probably not nutrient-limited. Nutrient transport capacity develops rapidly with nutrient stress (e.g. Flynn et al., 1999), when substrate limitation is important, and this compounds problems of measuring half-saturation constants (Flynn, 1998). Overall, these examples highlight the need for further investigation of the biological implications of the decreased diffusive conductance of the boundary layer around colonies in comparison with those around isolated unicells from colonies.
Larger colonies or organisms with flagella, such as Volvox, can have significant reductions in diffusive limitation on nutrient flux at the surface of the colony or organism as a result of advection by the flagella (Short et al., 2006; Solari et al., 2006a,b). Any restrictions on solute fluxes between the colony surface and the bulk medium would favour the retention of extracellular inorganic or organic storage compounds of low molecular mass, if these are truly storage (re-used) rather than simply accumulated compounds (Raven, 1982, 1997a). However, the possibility of long-term (more than a doubling time) storage of such compounds seems remote (Reynolds, 2007; but see Flynn & Gallon, 1990) in view of the known diffusivities of low molecular mass materials in the mucilage and the permeability of the pellicle of Phaeocystis to compounds at least as large as sucrose (Hamm et al., 1999). Flynn & Gallon (1990) present evidence consistent with short-term (over a light/dark cycle) storage of free amino acids in mucilage surrounding the unicellular diazotrophic cyanobacterium Gloeothece. We shall see later that the potential for storage of low relative molecular mass (Mr) compounds is greater in vacuoles, which are found predominantly in larger cells (Raven, 1995b; Menden-Deuer & Lessard, 2000), because vacuoles are much less intrinsically leaky (flux of solute per unit area per unit driving force acting on the solute) than are extracellular matrices, or pellicles.
The modelling by Yoshiyama & Klausmeier (2008) of optimal cell sizes for optimal nutrient uptake by planktonic micro-organisms integrates external diffusion, membrane transport and intracellular assimilation in relation to the size of resource molecules. As Yoshiyama & Klausmeier (2008, p. 68) point out, the application of this model to large phytoplankton unicells requires explicit consideration of the large fraction of the volume occupied by vacuoles in these organisms, which impacts not only on the balance of the two transport components relative to the assimilation term in the model but also on intracellular transport, resource storage, vertical motion and the package effect (Raven, 1984, 1997a).
3. Package effect
The ‘package effect’ involves a decrease in the specific absorption coefficient of a pigment molecule when it is aggregated versus homogeneously distributed. Theory and observation indicate that photon absorption per unit pigment will decrease with cell volume in unicells for any fixed pigment concentration (Duysens, 1956; Agusti, 1991). The occurrence of cells in a colony (or multicellular organism) relative to the same cells occurring in isolation would be expected to increase both the package effect with regard to the absorption of PAR and UV (Kirk, 1975a,b, 1976, 1994; Raven, 1984, 1991; Garcia-Pichell, 1994) and the restriction on nutrient uptake from low external concentrations as a result of the greater thickness of the diffusion boundary layers around larger objects (Raven, 1989, 1998a). Other factors being equal, both of these effects would have the same impact on a colony as on a giant unicell or multicellular organism of the same size and shape (Kirk, 1975a,b, 1976, 1994; Raven, 1984, 1989, 1998a; Garcia-Pichell, 1994).
Theoretical predictions regarding the package effect are consistent with empirical data on the effects of organism size obtained for Microcystis in a comparison of colonies with isolated unicells (Ganf et al., 1989), and with empirical data on the effects of varied chromophore concentrations obtained for Phaeocystis in a comparison of colonies grown at different irradiances (Moisan & Mitchell, 1999). These studies specifically addressed absorption of PAR where, within limits, ‘more is better’. However, the same physical principles apply to the absorption of UV-B radiation (Raven, 1991; Garcia-Pichell, 1994) where ‘less is better’. Extracellular polysaccharides as part of colony structure, whether homogenously dispersed or in a pellicle, do not absorb significant UV-B, but could potentially facilitate the attachment of UV-B screening compounds.
In general, extracellular compounds are better able to screen the protoplast than intracellular compounds. Phaeocystis colonies appear to have extracellular as well as intracellular UV-B screening compounds (Riegger & Robison, 1997; Moisan & Mitchell, 2001). However, the effectiveness of the screening compounds in these colonies is not sufficient to make them less sensitive to UV-B than co-occurring unicellular diatoms (Riegger & Robison, 1997); overall effects of UV-B are not just a function of screening compounds, but also of tolerance and damage repair (see e.g. Heraud & Beardall, 2000).
Any differences in the size dependence of the package effect and intracellular pigment composition between unicellular and colonial organisms could contribute to differences in the size scaling of light-limited or photoinhibited photosynthesis and growth between these growth forms.
4. Vertical movement
Upward and downward movement of cells is affected by a variety of factors. Vertical movement is important for phytoplankton because it can influence access to light and nutrients, as well as the eventual loss by sinking of all nonmotile cells that are denser than the medium. All else being equal, a larger object is expected to move more rapidly through a medium than a smaller one (formalized in Stokes’ Law). Although this is a complex area to investigate experimentally (Walsby & Holland, 2006), it is pertinent to our consideration of coloniality, as it means we might expect a colony to move more rapidly than its constituent cells.
Many other properties of phytoplankton can also affect their vertical movements. For nonflagellate phytoplankton, downward movement is favoured by increasing the ‘ballast’ through storage of carbohydrate (Boyd & Gradmann, 2002) or mineralization. Upward movement in some cyanobacteria is achieved using gas vesicles, with downward movement occurring when this buoyancy is more than offset by polysaccharide ballast (Walsby, 1994). Phaeocystis colonies have been observed to be buoyant in seawater (Peperzak et al., 2003) by an unknown mechanism, presumably involving manipulation of solute content (Boyd & Gradmann, 2002) as eukaryotes lack gas vesicles.
In freshwater eukaryotes, the capacity to rise relative to the immediate environment probably cannot be achieved by decreasing density by either changing intracellular (vacuolar) solutes or decreasing the ‘ballast’ provided from storage of carbohydrate (see Boyd & Gradmann, 2002). Positive buoyancy cannot readily be achieved by accumulating lipids even as a large fraction of the biomass, although this can occur in some strains of the green alga Botryococccus with up to 75% hydrocarbon in the dry weight (Young et al., 2004). The ‘genus’Botryococcus mainly contains strains from Trebouxiophyceae, although two strains occur in different clades within the Chlorophyceae (Serousy et al., 2004).
The capacity to move upwards relative to the surrounding water can be attained, and is certainly most readily controlled, by flagellar motility. Small freshwater colonies and unicells of the size of individual cells of Volvox can only swim at up to c. 150 µm s−1 (e.g. the freshwater Cryptomonas phaseolus has an ESD of c. 10 µm; Pedrós-Alióet al., 1987). However, large Volvox colonies of > 1 mm ESD can swim at speeds of > 1 mm s−1 as a result of decreased friction relative to the available energy per unit volume to power flagella. Limits on this available energy for a spherical structure come from the necessarily superficial disposition of the flagellate cells and the package effect, which limits the possibility of increased PAR absorption by the light-harvesting apparatus as fractional absorption approaches 1.0 (Raven, 1984).
The ecological and evolutionary relevance of possible cyclic upward and downward movement by phytoplankton seems to relate to inverse gradients of PAR (highest at the surface) and nutrients such as nitrogen (N) and phosphorus (P) (highest at depth), combined with a small vertical component of water movement in the upper mixed layer/epilimnion. These ecological conditions are met, at least as far as gradients are concerned (see modelling by Flynn & Fasham, 2002), for very large unicellular diatoms and colonies of large diatom cells (e.g. species of Coscinodiscus, Ethmodiscus and Rhizosolenia), prasinophyte phycomata, large dinoflagellates in parts of the oligotrophic ocean, dinoflagellates in fjords, and Volvox in lakes (Swift & Meunier, 1976; Kahn & Swift, 1978; Raven & Richardson, 1984; Kirk, 1998; Raven, 1998b). Migrations generally occur on a diel basis (near the surface during the day; deeper at night), but can take several days in the oligotrophic ocean. The vertical migration of Microcystis has been modelled by Rabouille et al. (2005) and Rabouille & Salençon (2005), and that of the dinoflagellate Alexandrium by Flynn (2002). For Trichodesmium in an oligotrophic region of the north Pacific, modelling suggests the possibility of migrations between the upper euphotic zone and the phosphocline, but only for colonies above a threshold size (White et al., 2006). However, it must be remembered that these vertically migrating phytoplankton organisms share their habitat with photolithotrophs that do not show such migration and also with potential predators that often migrate upwards during darkness: as usual in ecology, one size does not fit all.
Ecological interpretation of these different characteristics of phytoplankton organisms is complicated by properties of the physical environment. For example, turbulence is a determinant of the rate of sinking. Recent findings indicate that turbulence can increase the average settling velocity of phytoplankton cells, rather than decreasing it, as had earlier been believed (Ruiz et al., 2004), although the generality of these findings remains to be tested there may also be more rapid formation of cell aggregates under turbulent conditions; aggregates sediment very much more rapidly than single cells (Jackson, 1990). Also, the mixing depth affects the distribution of nutrients in the water column, and the probability of cells migrating to the nutricline before they become light (energy) limited (see simulations in Flynn, 2002).
5. Turgor pressure and turgor-resisting cell walls
Some attributes of colonial and multicellular algae are shared to varying degrees by large, vacuolate unicells. Here the thicker cell walls (Raven, 1982, 1995a) of those with turgid cells replace the extracellular materials within the colony as having the greater potential for increased screening of UV-B, compared with individual small cells. The only photosynthetic cells of this kind are eukaryotic (e.g. diatoms, dinoflagellates, prasinophytes and phycomata), although some sulphide-oxidizing chemolithotrophic bacteria have ‘eukaryotic’ nongaseous vacuoles (Raven, 1982, 1997a). According to the Laplace relationship, turgid eukaryotic cells devote the same fraction of their volume to cell wall material, regardless of cell size, given a particular cell wall material, turgor pressure and mechanical safety factor (Raven, 1982, 1995a). This relationship applies also to the turgor-resisting peptidoglycan layer of cyanobacterial cell walls (Hoiczyk & Hansel, 2000). The nonturgor-resistant theca of dinoflagellates was not found to significantly increase the carbon (C) per unit volume when thecate and athecate species were compared over a large size range (Menden-Deuer & Lessard, 2000). Villareal et al. (1999) showed that the previously determined relationship between silicon (Si) content and cell volume for diatoms up to a volume 106 µm3 applied to cells up to a volume of 109 µm3, so that there is a tight linear relationship between cell volume and cell Si content, as predicted by the Laplace relationship. These relationships apply to cells growing under near-optimal resource supply conditions; under resource limitation the Si content is increased (see Raven & Waite, 2004). For a given fraction of the maximum growth rate, the ballast effect of SiO2 is the same for a large unicell as for a colony of the same volume. However, because larger cells have a larger fraction of the cell volume occupied by vacuole than do smaller cells (Raven, 1995b; Menden-Deuer & Lessard, 2000), any moderating effect of changed vacuolar composition (Boyd & Gradmann, 2002; Raven & Waite, 2004) on overall cell density would be greater in a large unicell than in a colony of the same volume, assuming no greater contribution of extracellular polysaccharide, which is not part of the regular cell wall, in colonies than in unicells. Even vacuolate diatoms that might be expected to have adequate storage volume for dissolved low Mr solutes store high-density polyphosphate (Diaz et al., 2008). In addition, the amount of Si deposited in the diatom cell wall increases markedly with non-Si induced growth limitation (Martin-Jézéquel et al., 2000).
However, large three-dimensional colonies typically have less than half of their volume occupied by cells, so that there is less wall material per unit colony volume than there is in a cell of the same size. Particularly in cyanobacteria, the rest of the volume contains extracellular polysaccharides other than turgor-resisting cell wall material; this extracellular material can bear UV-B-screening compounds such as scytonemin (Proteau et al., 1993). The question of a turgor-resisting wall does not arise in cells with flagellar motility, as the necessarily wall-less flagella are in hydraulic connection with the main cell (Raven, 1982, 1995a).