The biotic and abiotic factors that govern the productivity of natural phytoplankton communities are well established: nutrients, light, temperature and losses to grazers. Fertilisation with inorganic nutrients is a sure route to establishing dense algal populations; however, the identity of the primary growth-limiting nutrient(s) remains a source of vigorous debate in aquatic ecology (Conley et al. 2009; Harpole et al. 2011). While excess phosphorus loading is a major cause of eutrophication of surface waters (Carpenter et al. 1998), the role of nitrogen is contentious (Smith 2003; Schindler et al. 2008; Conley et al. 2009). The importance of inorganic nitrogen supply to algal production depends on the favourability of conditions required for N2 fixation (Conley et al. 2009) and denitrification (McCrackin & Elser 2010). Co-limitation by multiple resources can arise through physiological or ecological mechanisms, and has been demonstrated frequently in fertilisation experiments with both aquatic and terrestrial autotrophs (Harpole et al. 2011). Light becomes limiting due to self-shading under eutrophic conditions where nutrients are plentiful and algal biomass is high, resulting in reduced nutrient uptake efficiency by algal cells. Thus, the resources that potentially limit algal biomass production can vary greatly depending on local environmental conditions.
In addition to influencing the production of algal biomass, the relative supplies of different mineral nutrients and light can have strong effects on the species composition, elemental stoichiometry and biochemistry of phytoplankton, and therefore their value as biofuel feedstocks. Large supplies of P relative to N may favour competitive dominance by heterocystous N2-fixing cyanobacteria (Smith 1983; Schindler et al. 2008), which have low cellular lipid contents (Griffiths & Harrison 2009; Rodolfi et al. 2009). However, a range of other environmental factors including salinity (Conley et al. 2009), temperature (Johnk et al. 2008; Paerl & Huisman 2008) and light (Pinto & Litchman 2010) determine whether N2-fixers become prevalent in phytoplankton communities at any given N : P supply ratio. In addition, many but not all species of eukaryotic algae accumulate lipid in their cells under conditions of N-starvation when carbon is primarily used to produce either starches or lipids instead of proteins (Spoehr & Milner 1949; Rodolfi et al. 2009). This intraspecific variation in cellular lipids represents a trade-off between optimising the yields of total algal biomass vs. lipids in large-scale production facilities designed to harvest lipids for downstream transesterification into liquid biodiesel.
Productivity patterns in natural ecosystems indicate that fertilisation strategies need to be carefully considered to avoid excess nutrient loading that can contribute to downstream eutrophication, shift the competitive balance between targeted algal crop species and invasive algal weeds, and maintain both rapid growth and a favourable lipid content of algal biomass. Water from domestic wastewater treatment plants or livestock operations offers a potential source of cheap nutrients, and also provides the ancillary benefit of recycling of nutrients that might otherwise be discharged into surface waters (Craggs et al. 2011; Sturm et al. 2012). Wastewater or sea water are likely to be the only viable sources for growing algal biofuels in many of the arid, high irradiation regions that could offer the highest production without displacing arable land. However, the variable water chemistry and microbial communities contained in these water supplies may preclude the possibility for maintaining monocultures of selected or engineered algal lineages with desirable properties for biofuels as sterilisation of the inflowing water would be infeasible and cost prohibitive.
Estimating the nutrient demands of large-scale algal biofuel production
The resource requirements of algal biofuel production have a major impact on the economics and sustainability of biomass cultivation systems (DOE 2010; Georgianna & Mayfield 2012). Large quantities of inorganic nutrients will be essential to support billion gallon per year-levels of liquid biofuel productivity. However, the life cycle impacts of these resource inputs on algal biomass cultivation have only recently been considered (Clarens et al. 2010; Pate et al. 2011).
Our knowledge of nutrient physiology and ecological stoichiometry can help inform estimates of the nutrient supply rates that will be needed to support commercial-scale algal biofuels. Phytoplankton use photosynthesis to convert light energy into algal biomass that typically has the following elemental stoichiometry (Stumm & Morgan 1981):
The typical carbon (C) to nitrogen (N) to phosphorus (P) stoichiometry of C106 : N16 : P1 by moles for algal biomass is commonly referred to as the Redfield ratio (Redfield 1958). Although the Redfield ratio is not universal, this ‘average’ cellular stoichiometry allows quantitative predictions to be made about the quantities of carbon, nitrogen and phosphorus required for algal production. Pate et al. (2011) used Redfield C : N : P stoichiometry to estimate the nutrient requirements of future commercial-scale algal biodiesel production. They calculated the N and P required to produce one metric tonne of dry algal biomass as follows. Converted to a dry mass basis, the C : N : P ratio in algae becomes
Assuming that cellular carbon is on average 50% of algal dry weight, Pate et al. (2011) concluded that each metric tonne of dry algal biomass produced requires 88 kg of elemental N and 12 kg of elemental P if it adheres to Redfield ratio stoichiometry. Their conclusions were pessimistic for the future commercialisation of algae biofuels: the projected nutrient requirements needed to support their lowest scenario target production level (10 billion gallons of algal bio-oil feedstock per year, assuming 50% oil content in the harvested biomass) represented 44% of the total US consumption of N from ammonia, and 20% of the total US consumption of P from phosphate rock, in 2006.
The above calculations are likely to be sensitive to the potential to recycle N and P contained in algal biomass after extracting lipids, and to significant deviations from Redfield stoichiometry that have been observed (Sterner et al. 2008). The Redfield Ratio characterises diverse natural algal assemblages, however, and the species that make up those communities can vary considerably in elemental composition (Geider & La Roche 2002). Stoichiometry at the community level is also variable within and among habitats. Sterner et al. (2008) analysed more than 2000 measurements of the chemical content of suspended particulate matter from freshwater and marine ecosystems worldwide, and found that a non-Redfield stoichiometry of C166 : N20 : P1 by moles best described the elemental composition of algae. This conclusion implies a significantly higher average nutrient use efficiency for algal biomass production (NUE, defined here as the number of moles of carbon fixed into algal biomass per mole of cellular N or P) than predicted by the Redfield ratio. Algal communities in small freshwater lakes, which should resemble the assemblages that would develop over time in open, freshwater raceway pond bioreactors exposed to invasions by wild algae, exhibited even higher average elemental ratios of C : P = 224 by moles and C : N = 10 by moles. These values imply that higher nitrogen- and phosphorus-use efficiencies could potentially be achieved in algal biomass production systems than assumed based on the Redfield ratio (Pate et al. 2011). However, the slopes of the lines relating C content to both N and P content of seston were < 1, indicating that the most productive water bodies (which likely bear closer resemblance to highly productive bioenergy communities) had stoichiometry closer to Redfield ratios than oligotrophic lakes.
Recent stoichiometric data from pilot-scale outdoor bioreactors indicate greater P demands but lower N requirements for biofuel production than predicted by Redfield ratios. Sturm et al. (2012) performed a pilot-scale algal production experiment using four bioreactors with pre-chlorination effluent withdrawn from the final clarifier of the Lawrence, KS, domestic wastewater treatment plant. Each bioreactor had a volume of 10 m3 that was continuously aerated without CO2 supplementation and a hydraulic residence time of 10 days after being seeded with a diverse plankton inoculum from a nearby lake. The molar stoichiometry of the dry algal biomass produced by these outdoor bioreactors was C67 : N3.9 : P1, indicating far lower phosphorus-use efficiency than predicted either from the classical Redfield stoichiometry or from Sterner et al.'s (2008) average of C : P = 224 (Fig. 1a). In contrast, the biomass produced contained a much higher average C : N ratio of 17 : 1 by moles and thus a much greater nitrogen-use efficiency than predicted either by Redfield (C : N = 6.6 : 1) or Sterner et al. (C : N = 10 : 1, Fig. 1b). Further research is required to determine the appropriate C : N and C : P ratios for calculating the nitrogen and phosphorus demands of large-scale algal biofuel efforts, and must take into account large deviations from Redfield elemental stoichiometry that have been identified.
Figure 1. Ranked values of (a) algal carbon:phosphorus stoichiometry (C : P ratios, by moles) and (b) algal carbon:nitrogen (C : N) in natural freshwater systems from the database of Sterner et al. (2008).
Download figure to PowerPoint
Top-down control of production
Many of the same traits that make algae highly productive also increase their susceptibility to a wide range of herbivores. Consumers including micro- and meso-zooplankton exert strong control over algal production and density (Brooks & Dodson 1965), influencing seasonal dynamics of production (Sommer 1989). Heterotrophs appropriate on average ca. 50% of net primary production in phytoplankton communities, a 3× greater fraction than their counterparts in terrestrial ecosystems (Cebrian 1999). The stronger top-down control of aquatic production may arise from the closer stoichiometric agreement between phytoplankton, which have low C : N and C : P ratios in their cells, and their heterotrophic consumers (Elser et al. 2000; Cebrian et al. 2009). The greater productivity of biofuels by phytoplankton relative to terrestrial crop plants therefore comes at a cost of greater susceptibility to pest outbreaks and crashes due to grazing.
In addition to being more effective herbivores, the consumers of algae also differ from many terrestrial pests in being highly dispersive and cosmopolitan in their distribution (Shurin 2000; Finlay 2002). Unicellular and metazoan heterotrophs rapidly colonise even very small water bodies (Jenkins & Buikema 1998; Cáceres & Soluk 2002; Cohen & Shurin 2003). Park et al. (2011) noted that attempts to grow algal monocultures in open ponds for periods longer than three months all failed due to contamination by wild algae and/or zooplankton. Preventing contamination by consumers in large, open outdoor biofuel ponds will therefore likely rely more on limiting the success and impact of colonists than preventing their dispersal. Biocontrol schemes invoking trophic cascades from fish predators have been proposed (Smith et al. 2010; Kazamia et al. 2012) and may prove effective for large crustacean grazers. However, more specialised microbial pathogens like fungi (Ibelings et al. 2004), viruses (Suttle 2005) or prokaryotes (Kang et al. 2005) can exert substantial control over algal productivity, especially in low diversity systems. These consumers may not be subject to strong top-down control via vertebrate predators, although some such as chytrids may be vulnerable to grazing by crustacean zooplankton (Kagami et al. 2004; Hamilton et al. 2012). The impacts of microconsumers of algae may be better minimised by management strategies that manipulate biomass harvesting strategy, water chemistry or utilise mixed-species algal assemblages.
Test raceway ponds at Sapphire Energy's industrial test facility in Las Cruces, NM, USA (Fig. 2) illustrate the vulnerability of cultivated algae to invasion by contaminating microorganisms. Fig. 3 shows fluctuations in the density of one cultivated algal strain (Scenedesmus sp.) and pathogenic fungi (chytrids) detected by quantitative PCR in three replicate ponds. Densities of chytrids (as measured by ‘cycle threshold’, the number of PCR amplification cycles necessary to detect a given sequence) increased in concert with algae until application of a chemical fungicide on day 33 of the trial (indicated by the arrow) resulted in a sharp drop in chytrid abundance. The dynamics closely resemble those of wild chytrids which have been shown to terminate blooms of their diatom hosts in lakes (Ibelings et al. 2004; Gsell et al. 2013). Chytrids are a ubiquitous component of freshwater ecosystems, but their role in aquatic food webs and productivity is largely unknown (Kagami et al. 2007). Chytrids display considerable variation in host ranges, with some infecting many hosts and others restricted to individual species, strains or even host life cycle stage (Ibelings et al. 2004; Kagami et al. 2007). Biofuel monocultures may be more susceptible to epidemics than diverse communities in the same way as other agricultural crops, and disease spread has been shown to be slowed in the presence of genetically diverse cultivars (Tooker & Frank 2012). Fig. 3 illustrates some of the variability in the dynamics of algae and their pathogens. While the patterns of population abundance cannot conclusively implicate chytrids as controlling factors of algal density, they show the vulnerability of biofuel ponds to contamination by wild pathogens and indicate that crop protection needs to be a major research priority for algal bioenergy.
Figure 2. (a) shows an aerial view of the Sapphire Energy field test site in Las Cruces, New Mexico. (b) shows some of the 30.5 m raceways from which the data in Fig. 3 were generated.
Download figure to PowerPoint
Figure 3. Dynamics of a cultivated micro-alga (dry weight in g/L, solid points and lines, left y-axis) and pathogenic chytrid fungi (CT, cycle threshold for PCR detection, x-symbols and dashed lines) in three replicate raceway ponds (~ 130 m2 surface area) at Sapphire Energy. Chytrids were controlled with a chemical fungicide introduced on the date indicated by arrow. Dry weight of algae indicated by solid lines and circles, indicated on the left y-axis. The ‘x’ symbols and dashed lines indicate chytrid CT values (shown on the right y-axis). The General Additive Models shown by the lines are all significant (P < 0.05).
Download figure to PowerPoint
What determines the biochemical composition of algae?
Algae exhibit tremendous variation in elemental and biochemical composition both within and between species, and in response to environmental conditions. This plasticity in composition is an important aspect of their use as biofuel feed stock. Under favourable growth conditions, algae synthesise polar lipids for structural and signalling functions which reside mainly in the plasma membrane and organelles (Solovchenko 2012). In stressful environmental conditions, particularly under N-starvation, however, many microalgae activate neutral lipid biosynthetic pathways towards the formation and accumulation of neutral lipids, especially triacylglycerols (TAG, Hu et al. 2008). Cells exposed to nitrogen limitation often decrease in protein content and increase carbohydrate and/or lipid storage (Mandal & Mallick 2009; Feng et al. 2011; Jiang et al. 2012). Phosphorus limitation has also been found to enhance lipid accumulation in Nannochloropsis sp. (Rodolfi et al. 2009) and Monodus subterraneus (Khozin-Goldberg & Cohen 2006), as has stress from high salinity (Takagi et al. 2006), low pH and high light intensity (Damiani et al. 2010). Thus, while nitrogen starvation seems to be a general mechanism for inducing neutral lipid production across species, other environmental stresses also affect biochemical composition in idiosyncratic and species-specific ways.
Growth conditions also affect the chemical properties of TAGs in ways that influence their quality as a liquid fuel feedstock. For instance, the fuel properties of biodiesel (which is produced by the transesterification of TAGs) are determined by chain length and degree of unsaturation of fatty acid esters (Knothe 2005). Biodiesel derived from TAGs with high amounts of un-saturated fatty acids shows better lubricity and cold-temperature flow properties. On the other hand, increasing oxidative stability and ignition quality (as measured by the cetane number) requires decreasing the relative content of unsaturated fatty acids. Fuel quality is therefore optimised at intermediate levels of saturation. A good compromise can be reached by generating fuel high in the mono-unsaturated fatty acids, such as oleate or palmitoleate, and low in both saturated and poly-unsaturated fatty acids (Durrett et al. 2008). Several aspects of the algal cultivation environment affect the degree of lipid saturation. For example, low temperatures generally induce greater unsaturation in fatty acids in microalgae (Khotimchenko & Yakovleva 2005), while nitrogen deficiency results in the biosynthesis of storage TAGs enriched in saturated and mono-unsaturated fatty acids (Mandal & Mallick 2011). The quality of biodiesel produced therefore can be manipulated through the growth conditions imposed during algal cultivation; however, less is known about how other properties such as algal population growth rate and susceptibility to grazing may influence biofuel quality.
As in most organisms, traits of phytoplankton that determine fitness are not independent, but rather are often correlated. These correlations often represent trade-offs arising from mechanistic constraints based on principles of cell biology (Litchman et al. 2007). For example, cell size and surface area to volume ratio (SA/V) control major fundamental functions in phytoplankton. Small-celled microalgae have higher SA/V and more efficient nutrient utilisation and, thus, may be better nutrient competitors than larger cells (Chisholm 1992). However, storage capabilities, including lipid storage, are higher in large-celled microalgae (Litchman et al. 2009). Growth rate is unimodally related to cell size, peaking around 102μm3 in volume (Marañon et al. 2013). High growth rates also appear to be correlated with low lipid production (Smith et al. 2010), and, therefore, fast-growing microalgae may not be the best candidates for biofuel production. Trade-offs may also occur in competitive ability for different nutrients, e.g. nitrogen and phosphorus, or even in multidimensional trait space between competitive abilities for different nutrients and grazer resistance (Edwards et al. 2011). Light is also an essential resource and its utilisation is governed by physiological trade-offs. Algal species utilise different parts of the wavelength spectrum of light depending on the presence of particular accessory photosynthetic pigments (Stomp et al. 2004). If the same cell cannot produce all accessory pigments, then trade-offs may occur in the ability to harvest different wavelengths. These kinds of trade-offs may limit the potential to engineer or select algal species for biofuel production that maximise a desired function, such as lipid content, growth rate or photosynthetic efficiency. Such trade-offs have major implications for the selection of strains, and for the design of growing conditions and harvesting technologies. Understanding trade-offs is critical to maximising the performance of cultivated strains in terms of multiple functions, including growth, lipid concentration, nutrient demands and resistance to enemies.
Can genetic modification overcome trade-offs?
The development of genetically modified algae for commercial biofuel production is still in its infancy but proceeding rapidly (Georgianna & Mayfield 2012), and, to our knowledge, none has been approved for outdoor cultivation. Just as modern agricultural crops emerged from their wild progenitors, first through long-term selection and then by advanced breeding methods, the development of algae for commercially viable biofuel production depends on research to develop effective crop enhancement strategies. However, scientists are rapidly improving their ability to programme the behaviour of microbes for specific applications including biofuel production (Brenner et al. 2008; Kilian et al. 2011). A growing capacity to sequence genomes is also facilitating the identification, cloning and manipulation of genes, and also bringing the power of translational tools to bear in algal biotechnology. Much of this research is concentrated upon manipulating their genetics and cellular metabolism to increase cellular lipid concentrations, increase growth rates or confer biotic resistance against consumers. For instance, Simkovsky et al. (2012) identified the genetic basis for resistance to protozoan grazing in a filamentous cyanobacteria and proposed its use as a basis for crop protection strategies. Rosenberg et al. (2008) concluded that metabolic engineering may be necessary in order for microalgae to achieve commercialisation as biofuel crops. Gressel (2008) outlined seven targets for genetic modification to improve the performance of algal biofuel crops, including the acquisition of herbicide resistance, simultaneously maximising growth and lipid content, improving the efficiency of photosynthesis and CO2 utilisation, and improving strain biosafety by engineering an inability to survive in natural ecosystems.
We suggest that faith in the ability of genetic modification to overcome all biological limitations on algal biofuel productivity is inconsistent with the concept of trade-offs. For instance, it may not be feasible to engineer microalgae to simultaneously resist herbivore grazing and also be ecologically non-competitive to survive if accidentally released into the wild, or to simultaneously achieve high lipid yields and low cellular nutrient demands. In addition, microalgae are likely to evolve resistance to herbicides at a much faster rate than terrestrial crops because they have orders of magnitude shorter generation times (ca. 1–2 days, vs. 1 year) and larger population sizes. Bull & Collins (2012) outline several evolutionary challenges that will likely be associated with the use of algae as biofuel feedstock. They note that algae are grown as a continuously reproducing population and harvested periodically to maintain high yields, in a manner more similar to managed wild fish or game populations than terrestrial crop plants. They conclude that this propagation and harvesting method subjects cultivated algae populations to strong selection pressures which, in many cases, may produce traits that are at odds with maintaining algae suited for biofuel production. In particular, there is the strong possibility of selection for weedy mutants that grow fast or escape harvesting, as well as genetic drift and loss of selected or engineered traits. We agree with Bull & Collins (2012) that a clear understanding of evolutionary principles are needed to design algal growth and harvesting methods that reduce the negative impact of selection.
We also agree with Snow & Smith (2012) that although genetically modified microalgae are unlikely to persist if accidentally released into natural ecosystems, thorough ecological and evolutionary assessments nonetheless will be needed to test this assumption. Invasive microbes and their potentially transformative effects on invaded ecosystems are often cryptic and almost certainly impossible to reverse (Litchman 2010). The benefits of genetic modification of algae for biofuel purposes therefore must be weighed against environmental risks which, although likely to be small, are virtually unknown at present.
Diversity & productivity
The conclusion that species diversity promotes ecosystem productivity and stability has become canonised as one of the basic tenets of ecology (Cardinale et al. 2011). This principle has been applied to biofuel production by grassland plants, which produce similar amounts of energy as crops such as corn or soybean crops without inputs of fertilisers, pesticides or water (Tilman et al. 2006). In addition, Stockenreiter et al. (2012) showed that both natural and constructed algal assemblages of high diversity produced more biomass and more total neutral lipids than those containing fewer species (see also Corcoran & Boeing 2012). These studies provide encouraging indications that the use of polycultures may be an effective approach to enhancing algal bioenergy production.
Diversity can enhance productivity either through sampling effects in which diverse communities are more likely to include the most productive species under any set of conditions, or through niche differentiation and complementarity in resource use (Loreau & Hector 2001). Transgressive overyielding occurs when polycultures produce more biomass than monocultures of any of their constituent taxa, and results from facilitation or complementarity among species that allows a diverse community to acquire resources and convert them to biomass more efficiently (Fox 2005). Non-transgressive overyielding is the outcome of sampling effects and represents situations where polycultures are more productive on average than monocultures but do not out-produce the best single species. The distinction between overyielding and the sampling effect is important in the context of biofuels because benefits are only derived from co-culturing multiple taxa when polycultures out-perform their respective monocultures. If the sampling effect is most important, as has been shown in a meta-analysis of biodiversity–ecosystem function experiments (Cardinale et al. 2006), then monocultures can produce equivalent yield to polycultures if the most productive monoculture can be identified a priori.
Although polycultures may fail to out-perform highly productive monocultures in tests of algal diversity effects on biofuel productivity, there may still be some reason to expect benefits to be realised from co-culturing diverse taxa. First, the best performing single taxon may be difficult to pinpoint in advance, given inherent environmental uncertainty in weather and water chemistry. Diversity may reduce temporal variability in productivity through the portfolio effect (Doak et al. 1998) by increasing the likelihood that a productive species will be present under prevailing conditions. In this scenario, diversity acts as a form of insurance against unavoidable environmental stochasticity. Second, species-rich communities may also be less susceptible than monocultures to the effects of top-down control via grazers or pathogens. Because algal taxa vary greatly in their quality as food resources for different consumers, a diverse assemblage is more likely to contain inedible taxa that can resist many of the grazers that are likely to invade biofuel ponds (Duffy 2002). In addition, the presence of inedible prey can indirectly lower the attack rate on vulnerable prey by reducing the consumer's foraging rate and increasing handling time. Kratina et al. (2007) showed that the presence of inedible species in protozoan communities dramatically decreased the attack rate of predators on their preferred prey. Hillebrand & Cardinale (2004) showed evidence for these effects in a synthesis of experiments where consumers of benthic algae exerted stronger top-down control on low diversity periphyton assemblages. This principle might be applicable to algal biofuel communities if diverse algae decrease the likelihood of population crashes due to invasion by wild grazers.
We performed a laboratory experiment to test the idea that algal diversity enhances biomass production and nutrient uptake, and invasion by grazing zooplankton. We cultured communities consisting of 1, 2 or 5 species of phytoplankton (belonging to the groups Chlorophyta, Cyanophyta, Bacillariophyta and Heterokontophyta) with 10 unique combinations at each richness level drawn randomly from a pool of 10 species, as well as the full community of all 10 species (Fig. 4). These 31 unique diversity and composition treatments were crossed with the addition of Daphnia pulex grazers on day 25 of the experiment, and each treatment was replicated three times in 200-mL batch cultures. A full description of the methods of the experiment is given in the Online supporting information.
Figure 4. The results of a laboratory experiment growing multiple combinations of 1, 2, 5 and 10 species of microalgae. Each box on the x-axis of the left panels represents a unique microalgal species combination which was assigned by random draws from the full 10 species pool. The right panels show the distributions for each richness level. Rows show total algal biovolume (top), dissolved phosphate concentration in the medium (middle) and daily survival rate of Daphnia grazers on day 29 of the experiment (bottom). ANOVAs with species richness, Daphnia, and species combination nested within richness as factors found that both richness and species combination had significant effects on biovolume and PO4 concentration (P < 0.0001), but that addition of Daphnia had no effect. Richness, but not species combination, also had a significant effect on Daphnia survival (P = 0.02). The codes for the species making up the communities are shown on Fig. 5.
Download figure to PowerPoint
We found that mean total community biovolume increased with the number of species in a pattern most consistent with sampling effects. The two, five and 10 species polycultures mostly had final biovolume on par with, but not greater than, that of the most productive monocultures (Fig. 4). Fig. 5 shows the loge of the ratio of final biovolume in the polycultures to both the mean (Fig. 5a) and maximum (Fig. 5b) of the biovolume yield of the component species in monoculture. On average, the polycultures yielded 28.7% more biovolume than the mean of the component species in monoculture, but 19.9% less than the most productive species alone. However, one exception was a five species combination consisting of two cyanobacteria and three chlorophytes (mixture ABEHG in Figs 4 and 5, consisting of Aphanothece sp., Synechococcous elongatus, BL0910 (an unidentified green alga), Neochloris oleabundans and Chlorococcoum sp.). This combination yielded 15% more biovolume than any of the other polycultures or monocultures, and more than 2× as much as any of the five species when grown alone. This result indicates that overyielding occurs but is a relatively rare outcome, as has been shown in experiments with other groups of organisms (Cardinale et al. 2011). Identifying overyielding combinations of species is key to the application of community engineering to algal bioenergy.
Figure 5. Final biovolume yields of species mixtures relative to the average biovolume of the component species in monoculture (a) and the maximum of the component monocultures (b). Each point is the loge of the ratio of the polyculture yield to either the mean or the maximum biovolume of the species making up the mixture when in monoculture.
Download figure to PowerPoint
In addition, the species mixtures on average acquired more nutrients from their medium and resisted invasion by grazers better than monocultures. Dissolved phosphate concentration, a measure of free resources not absorbed from the environment by algae, declined with the number of species seeded. Finally, the daily survival of introduced Daphnia grazers declined markedly at high algal diversity even though the total biomass of algal food resources increased. Although Daphnia survived in many cultures, low survival in nearly all of the high-diversity polycultures prevented us from testing the effects of algal species richness on the grazing impact of the consumer. The results of this experiment support the contentions that algal diversity promotes both high biomass yield and dissolved P acquisition, while simultaneously providing some resistance against invasion by grazing consumers.
Although polycultures on average showed increased biovolume yield and dissolved phosphorus acquisition, and decreased grazer survival, the specific combinations of species that showed high values for one of these functions often did not necessarily show high values for the others. The treatment mean values for Daphnia survival were uncorrelated with either biovolume yield (n = 31, Pearson r = −0.30, P = 0.10), or dissolved P concentration (r = 0.18, P = 0.34), while biovolume and dissolved P were marginally negatively correlated (r = −0.35, P = 0.05). The weakness of the correlation between the three functions we measured indicates that polycultures that optimise one important function for bioenergy (e.g. biomass yield) may be suboptimal in terms of other crucial functions (e.g. grazing resistance). This finding suggests that the ‘multivariate dominance effect’ (Duffy et al. 2003), or the capacity of diverse assemblages to simultaneously increase rates of multiple ecosystem functions, may be weak in algae. The patterns agree with Gamfeldt et al. (2013) and Zavaleta et al. (2010) who show that some ecosystem services are negatively correlated with one another, preventing simultaneous optimisation of rates along multiple axes.
Although strong evidence supports the idea that species diversity promotes high ecosystem productivity (Cardinale et al. 2006), the utility of polycultures for enhancing productivity or stability of highly managed agricultural or bioenergy ecosystems remains largely untested. Using diversity as an operational approach to intensify yield in industrial-scale bioenergy requires much greater predictive understanding of the species combinations and environmental conditions that give rise to productive, efficient and robust assemblages. Different communities containing the same number of species show tremendous variation in biomass productivity depending on the identities of the species present (Fig. 4). The challenge in applying principles relating biodiversity and ecosystem function to industrial bioenergy is therefore not simply to understand the relationship between diversity and production. Rather, the important task is to identify combinations of species that are productive and robust against vagaries of the weather and the inevitable contamination by wild taxa.
Another reason that haphazardly assembled polycultures may not benefit biofuel yield is the potential for underyielding, the situation where combinations of species are less productive than the average monoculture. Although underyielding is a relatively uncommon outcome in published studies that manipulated diversity (Cardinale et al. 2006), a number of examples show that many mixtures perform poorly relative to their component monocultures. Schmidtke et al. (2010) found that among eight species of algae cultured in the laboratory, the fastest growing species competitively displaced slower growers, but produced less biomass at steady state, implying a trade-off between maximum exponential growth at low density and carrying capacity. Jiang (2007) found evidence of similar negative selection effects among bacteria where the taxa that dominated in polyculture were often less productive, resulting in underyielding. Thus, some species in diverse mixtures may have negative effects on community production. The data shown in Fig. 5 also show a number of examples where polycultures yielded less final biovolume than either the maximum or the average of the component species on their own (negative values of yield relative to the mean or maximum monoculture). Together, these results suggest that polyculture cannot be applied as a blind insurance policy because mixtures of species can yield either much more or much less biomass than any of their component monocultures.
Chemical ecology of algal aquaculture
Aquatic plants and animals produce a variety of biomolecules that play critical roles in almost every aspect of the functioning of aquatic communities and ecosystems. Compounds not involved in basic biochemical functions are referred to as secondary metabolites, and play critical roles in interactions within and between aquatic species. Phytoplankton secondary metabolites can suppress grazers and competitors (Hay 2009), and synchronise cell death during decline of phytoplankton blooms (Vidoudez & Pohnert 2008). An understanding of phytoplankton chemical ecology may have applications to the design and implementation of artificial algal communities for biofuel production and may provide valuable tools for crop protection.
Phytoplankton have evolved diverse morphological and chemical adaptations to grazing pressure (Long et al. 2007). Cyanobacteria, long recognised for their role in freshwater harmful algal blooms, produce structurally diverse secondary metabolites including microcystins that have toxic and inhibitory effects on zooplankton grazers (Wiegand & Pflugmacher 2005). Chemical defences have also been documented in other taxa. For example, copepod grazers showed reduced feeding activity and reproductive success when exposed to the toxic haptophyte Prymnesium parvum (Sopanen et al. 2006), strongly suggesting chemical mediation. Botryococcus braunii, a species of interest as a biofuel crop, produces high levels of hydrocarbon fatty acids in the form of triterpenes (up to 30–40% dry mass) that have suppressive effects on some zooplankton (Chiang et al. 2004). Harnessing these compounds in one form or another may be a viable approach to crop protection in biofuel aquaculture.
Incorporating chemically defended, herbivore-resistant phytoplankton into biofuel pond communities may confer crop protection through an ‘associational refuge’ for more herbivore-susceptible crop species. Similar strategies have been proven effective in mixed cultivars of plant crops (Tooker & Frank 2012). An associational refuge is an interaction where a susceptible species benefits from protection against grazing conferred by growing in association with a defended species. Associational refuges have been well documented in macroalgae (Duffy & Hay 1994; Levenbach 2008). To our knowledge, there are no well-documented cases of associational refuges in phytoplankton communities. However, compounds released by chemically defended phytoplankton have negative impacts on both zooplankton herbivores and phytoplankton competitors (Chiang et al. 2004; Ianora & Miralto 2010). Therefore, allelopathy and defensive chemistry, the main ingredients of associational refuges, occur in phytoplankton communities. Studies documenting the presence or absence of these interactions are ripe areas for future research in chemical ecology of phytoplankton communities. Defensive traits are the targets of many efforts to select or engineer defended strains of algae for bioenergy (Gressel 2008; Georgianna & Mayfield 2012; Simkovsky et al. 2012). However, it may not be necessary to incorporate defensive traits into the genome of crop taxa if their benefits can be realised by cultivating strains in the presence of defended taxa. The costs and benefits of such a strategy depends on the degree of crop protection conferred, the competitive interactions between the defended taxa and the crop, and the losses to grazers.