Introduction

Microalgae have been cultured for a long time because of their efficiency in converting solar energy into a great number of metabolites of commercial interest (Borowitzka 1999, 2013a). More recently, they have received increasing attention as a promising renewable source of carbohydrates and lipids for the production of renewable biofuels, as they can grow on non-agricultural land, using non-potable water, and produce oils much more efficiently than crop plants (Chisti 2007; Fon Sing et al. 2013). Although several species are being cultured commercially in large-scale outdoor facilities (e.g. Dunaliella salina in Australia and Israel and Spirulina in the USA, China and Thailand (Belay 1997; Borowitzka 2013b) very successfully for high value products such as β-carotene and food supplements, the challenge remains to reduce the capital and operational costs of algal culture.

Of the two commonly used large-scale microalgae culture systems (open ponds and closed photobioreactors), open systems are the main ones used commercially due to lower capital outlay for their construction and lower operational costs. But, in order to achieve reliable high productivities, good water circulation is required, as it reduces nutritional and thermal gradients and disrupts boundary layers, promoting nutrient availability to the cells, elimination of oxygen and reduces cell settling, anaerobiosis and biofilm formation (Persoone et al. 1980; Richmond 2013). Flow rates of 20–30 cm s−1 are most commonly used to achieve these aims (Borowitzka and Moheimani 2013). Higher rates of mixing consume too much energy, and lower velocities result in the previously mentioned unwanted outcomes (Oswald 1988; Lundquist et al. 2010). Mixing is not only one of the major parameters affecting algal growth but also represents a significant portion of the total annual power consumption and, therefore, annual operating costs (Stephens et al. 2010). Powering paddle wheels represents between 7.5 and 25.7 % of annual operating costs of production, depending on energy prices and facility design. Paddle wheel power consumption could be as high as 10.25 % of the cost of production of 1-L biodiesel for a 10-year return on investment (Rogers et al. 2014). Therefore, further optimization is absolutely necessary to determine the minimum mixing energy needed for an efficient growth at large-scale facilities.

Lundquist et al. (2010) determined that the power consumption of paddle wheel-driven ponds of 30-cm depth, operating continuously at 25 cm s−1 water channel velocity, would be 2.4 kW ha−1. If the mixing velocity can be slowed to 20 cm s−1 at night (10 h), when the mixing is not necessary to promote better photosynthetic activity, the energy consumption would be reduced to 17 % (Lundquist et al. 2010). Using the assumptions made by Lundquist et al. (2010), a far more significant power saving of 37 % could be achieved if paddle wheel operation were stopped between sunset and sunrise.

Photovoltaics assembled in solar panels are a well-known method for producing electrical energy from the sunlight and could be integrated into an algal culture facility to reduce power costs. The integration of photovoltaics with agricultural production in greenhouses has already been shown to be of benefit (Parida et al. 2011). This would be particularly useful for cheaper and more efficient production of bioenergy in remote locations away from sources of electrical power, avoiding the dependence on batteries and/or diesel fuel generators (Moheimani and Parlevliet 2013). However, if solar panels are the sole source of generated electricity for a regional raceway pond system, the paddle wheels will only operate during daylight hours.

Tetraselmis suecica has been largely grown in outdoor conditions for aquaculture purposes (Borowitzka 1997). Currently, it is being investigated for biofuel production in paddle wheel-driven raceway ponds. The long-term stability of T. suecica in open raceway ponds makes it a suitable candidate for outdoor large-scale cultivation (Fon Sing 2010).

Here, we studied the growth of T. suecica in open raceway ponds under three different mixing conditions. The paddle wheel of one raceway pond was operated continuously while the other two ponds were variants of daytime-only mixing; one pond was turned on an hour before sunrise and the other 1 h after sunrise. The aim of this study was to evaluate the impact on culture performance if mixing operations are stopped during night time, as would occur with paddle wheels driven by solar power only.

Materials and methods

Tetraselmis suecica CS-187 was grown in three identical fibreglass paddle wheel-driven raceway ponds with a 1-m2 surface area and depth of 16–20 cm (Raes et al. 2014). The culture of T. suecica CS-187 was obtained from the CSIRO Microalgae Research Centre (Hobart, Australia) and maintained in f/2 medium (Guillard 1975; Guillard and Ryther 1962) at a salinity of 3.3 % (w/v) NaCl. Stock culture was grown at 25 °C in a constant temperature room under 65–85 μmol photons m−2 s−1 provided by combination of cool white and day light fluorescence lights on a 12 h:12 h light/dark period. Initial inoculum was scaled up from two 150 mL liquid cultures to 1.5-L aerated and stirred Erlenmeyer flask, then to a 12-L stirred and aerated carboy and finally to an 80-L bag reactor without CO2 injection (Moheimani et al. 2011).

The three outdoor raceway ponds were chemically sanitized before the beginning of the experiment by filling with 6 % (v/v) sodium hypochlorite. After circulating this mixture for 60 min, it was left to sit overnight. The ponds were then completely drained and flushed 12 times with tap water. All outdoor raceway cultures were maintained with f/2 medium (Guillard 1975; Guillard and Ryther 1962). Chemically sterilized seawater was used for preparing the medium. It was sterilized by adding 200 ppm of 12.5 % (w/v) sodium hypochlorite to a 2200-L seawater tank. The tank was then aerated 30 L min−1 overnight, and 200 ppm of 985 g kg−1 sodium thiosulphate was added on the following day to neutralize the remaining hypochlorite. The tank was then aerated for 5 h prior to use.

The 80-L bag reactor was used to inoculate two raceway ponds, resulting in an initial cell concentration of 1.5–1.9 × 105 cells mL−1. When these two ponds reached a cell density of 4.3 × 105 cells mL−1, 66 L of each one was transferred to a third pond, resulting in cell concentrations of 2.3–2.6 × 105 cells mL−1 and 20-cm depth. The three raceway ponds were operated in batch mode for 1 week.

Between 30 Dec. 2013 and 10 Jan. 2014, all three ponds were cultured semi-continuously, mixed 24 h a day, under the exact same conditions (nutrient replenishment, pH, solar incidence, paddle wheel velocity, etc.), and harvested to restore the initial cell densities of 3 × 105 cells mL−1 every 2 days. The semi-continuous operating mode was maintained by draining between 20 and 40 % of pond and adding same volume of fresh f/2 medium. Under these conditions, the cultures were assumed to be nutrient replete and exhibited no characteristics of nutrient limitation.

Initially, the three ponds’ paddle wheels were operated at a speed of 28 ± 2 rpm, generating a flow rate of 20 ± 1 cm s−1. After this preliminary phase, one of them was operated in a continuously stirred condition and the other two were not mixed during night time. In one case of these two non-stirred-overnight conditions, the paddle wheel would start 1 h after sunrise, simulating the time needed for the solar panels to be fully operational, and it would stop 30 min before sunset, when solar panels would start to decrease generation of energy. In the other case, the paddle wheel started 1 h before sunrise and stopped 30 min before sunset. Commencing mixing 1 h prior to sunrise allowed the culture to be fully mixed at sunrise. This was done to investigate culture’s response if it is fully mixed at the time of sunrise, when photosynthetic efficiency starts to rise until noon (Kurzbaum et al. 2010; Sharon et al. 2009). The three modes of operation were then maintained until the experiment concluded.

Water temperature, pH, dissolved oxygen (DO) and salinity were measured on line and recorded continuously every 30 min, using 600R Series YSI Multiparameter Water Sondes. Irradiance and evaporation/rain data were obtained from the Bureau of Meteorology weather station (Australian Bureau of Meteorology 2014).

Growth parameters were measured every Monday, Wednesday and Friday (for cell density and biomass productivity) and once a week (for organic biomass and lipid analysis), using the methods described in Moheimani et al. (2013). Total biomass, as total dry weight (DW), was determined using the method of Zhu and Lee (1997). Filters containing 5 mL of sample were washed with 5 mL 0.65 M ammonium formate solution to remove salts. Total lipid was determined using the method of Kates and Volcani (1966) as adapted by Mercz (1994). Total carbohydrate was determined by using the method of Kochert (1978) as modified by Ben Amotz et al. (1985). Total protein content was determined using a modified Lowry et al. (1951) method based on Dorsey et al. (1978).

A Water-PAM (Walz GmbH) was used for chlorophyll-a fluorescence measurements during a 24-h follow-up experiment. This instrument uses LEDs for actinic/saturation (660-nm peak) and measuring (650-nm peak) light. Fluorescence measurements were conducted at 2 am, 5 am, 6:30 am, 8:30 am, 1 pm, 2:30 pm, 6:00 pm and 10 pm. Three samples were collected from each pond at each time period, with each sample being immediately transferred to the fluorometer and exposed to a short (2 s) pulse of far-red light followed immediately by a saturation pulse (0.8 s) to gain estimates of the minimum (Fo′) and maximum (Fm′) fluorescence in the light. The maximum PSII quantum yield in the light or Fv′/Fm′ was then calculated following Cosgrove and Borowitzka (2006). This parameter was used over effective quantum yield or (Fm′-F′)/Fm′ to negate some of the variability between samples due to unavoidable changes in the fast components of non-photochemical quenching during the sampling process.

Results

Initial results indicated that growth of T. suecica was identical in all ponds (Fig. 1 and Table 1). Average specific growth rate was 0.50 ± 0.05 day-1 (repeated measure one-way ANOVA, P > 0.05). Average yield was 288 ± 40 mg L−1 (repeated measure one-way ANOVA, P > 0.05), and average total dry weight productivity was 54 ± 6 mg L−1 day−1 (repeated measure one-way ANOVA, P > 0.05). At this point on 11 January 2014, with all three ponds at the identical growth stage, the manipulation of paddle wheel operation was commenced in two of the three ponds.

Fig. 1
figure 1

Solar irradiance, culture temperature and growth of Tetraselmis suecica (solid line) grown semi-continuously on three 1-m2 raceway ponds in Perth, Western Australia. Paddle wheels were operating a continuously, b from 1 h after sunrise until 30 min before sunset and c from 1 h before sunrise until 30 min before sunset. Contamination (broken line)

Full size image
Table 1 Comparison of growth parameters in different stages of Tetraselmis suecica along the experiment. Paddle wheels were operating (a) continuously, (b) from 1 h after sunrise until 30 min before sunset and (c) from 1 h before sunrise until 30 min before sunset. [Data are mean ± se (n = 5)]
Full size table

Over the first week of treatment, no statistically significant difference was found in growth parameters among the two non-mixed overnight ponds and the continuously mixed control pond (Fig. 1 and Table 1). However, on the ninth day, a sudden heat wave (over 35 °C in the pond) affected cultures and resulted in growth decline in all ponds. The rate of decline was the same across all ponds (μ = −0.57 ± 0.08 day−1; repeated measure one-way ANOVA P > 0.05) (Fig. 1 and Table 1). Within the first 24 h after the intense heat of midday 19 Jan, cells started to die. On the next day, severe morphology changes occurred, with bigger size and weight of cells and thicker cell walls. They did not divide until the fourth day of this short period, but they continued to grow in size and weight.

When weather conditions improved, T. suecica started to recover; however, recovery was not equal across all ponds. The continuously mixed pond recovered fastest, with a growth rate of 0.63 ± 0.04 day−1 in comparison with 0.42 ± 0.04 and 0.43 ± 0.04 day−1 in ponds ‘b’ and ‘c’, respectively (Table 1). No significant difference was found between the raceway pond mixed from 1 h before sunrise and the pond mixed from 1 h after sunrise for any of the growth parameters during the recovery phase (μ = 0.42 ± 0.03 day−1, productivity = 47 ± 0.06 mg L−1 day−1, repeated measure one-way ANOVA P > 0.05) (Fig. 1 and Table 1).

Although cultures became contaminated with protozoa after the first intense heat event, contaminant density never exceeded 1 × 104 cells mL−1. High frequency of harvest and high harvest volume as T. suecica growth recovered resulted in the removal of most of the contaminants from the ponds (Fig. 1).

Dissolved oxygen

Maximum oxygen concentrations in culture water of up to 12.2 mg mL−1 were recorded between 8:00 and 9:00 am every day. After this, dissolved oxygen (DO) in each pond would decrease until 4:30 am of the next day, when it would start to rise again until peak time. A slight increase was observed after harvesting.

All three ponds showed same daily pattern during the whole experiment, although a slight decrease in the DO profile was observed after the heat stress phase (Fig. 1). Maximum DO values during this phase were between 9.7 and 10.0 mg mL−1, with no significant difference among the three mixing conditions. Figure 2 shows average values over the first 5 days (left) and last 5 days (right) of the experiment.

Fig. 2
figure 2

Dissolved oxygen concentration and pH diurnal changes of T. suecica for the three different conditions. a Paddle wheels were operating: continuously (solid line), b from 1 h after sunrise until 30 min before sunset (dotted line) and c from 1 h before sunrise until 30 min before sunset (broken line) and irradiance (solid thick line). Average values of the first 5 days (left panel) and the last 5 days (right panel) of the experiment

Full size image

pH

Average pH values over the first 5 days (left) and last 5 days (right) of the experiment show a large and rapid increase in pH between 8:00 and 11:30 am, a decline toward 4:00 pm, with a dip at this time due to the impact of harvesting. Another pH increase occurred between 5:30 and 7:00 pm. Maximum pH differences between the continuously mixed pond and the other two ponds of 0.5 pH units were recorded during night in the first days of the experiment (Fig. 2, left panel).

The period of intense heat resulted in a reduced pH recorded at each time of the day. This slight acidification of the medium exhibited the same pattern in all three ponds, and the difference in pH during night between the continuously mixed pond and the other two ponds was diminished (Fig. 2, right panel).

Cell weight

During the period of senescence between 19 and 23 January, cell weight increased between three- and fivefold in all conditions compared to the average values recorded during the semi-continuous period leading up to the heat wave (Fig. 3 and Table 1). In fact, during the senescence period of the experiment, when a negative specific growth rate was observed, positive values of daily biomass productivity [mg L−1 day−1] were achieved due to the increase in cell weight (of the surviving cells) (see Fig. 3 and Table 1). Once the recovery phase started, the weight of the cells started to decrease in all ponds but remained greater than before this short 5-day period of extreme heat stress.

Fig. 3
figure 3

Changes in cell weight of T. suecica grown under three different mixing conditions along the duration of the experiment. From 11 Jan, paddle wheels were operating: a continuously (broken line, empty circle), b from 1 h after sunrise until 30 min before sunset and (broken line, inverted triangle) and c from 1 h before sunrise until 30 min before sunset (solid line, filled circle)

Full size image

Cell weight decreases during night time and starts to increase immediately after sunrise with a consistent peak in the hour following sunrise regardless of the mixing regime. This was also observed in the 24-h follow-up. Lipid, protein and carbohydrate contents show an inverse tendency, increasing during night time, although not enough data is available to reach statistically significant conclusions (Fig. 4).

Fig. 4
figure 4

Changes in growth parameters during the 24-h follow-up. Each panel shows the changes in growth parameters along the day in cultures of T. suecica grown under three different mixing conditions, meaning that paddle wheels were operating a continuously (filled circle), b from 1 h after sunrise until 30 min before sunset (empty circle) and c from 1 h before sunrise until 30 min before sunset (inverted triangle)

Full size image

Chlorophyll fluorescence

There were no significant differences in Fv′/Fm′ between the ponds (Fig. 5). Fv′/Fm′ dropped significantly in the hour after sunrise, increased slightly to ∼0.4 for the remainder of the day and recovered in the hour after sunset (Fig. 5).

Fig. 5
figure 5

Maximum PSII quantum yield of light-adapted T. suecica (Fv′/Fm′) and irradiance changes along the day during the 24-h follow-up. Paddle wheels were operating a continuously (filled square with solid line), b from 1 h after sunrise until 30 min before sunset (filled square with dotted line) and c from 1 h before sunrise until 30 min before sunset (filled square with broken line). Irradiance (dotted curve)

Full size image

Discussion

In favourable growth conditions, T. suecica biomass productivity was not affected by stopping mixing during night time. This was indicated by no significant difference in the photosynthetic performance (ANOVA repeated measure one way; P > 0.05), growth rate and biomass productivity of T. suecica in a continuously mixed raceway compared to raceways where mixing was stopped overnight.

While further investigations need to be done at a larger scale and during winter time, these findings may have significant cost-saving implications for large-scale culture of T. suecica and similar species. A number of studies have investigated manipulation of the mixing regime to realize cost savings. These have included reducing the mixing velocity during 10 h of the day (Lundquist et al. 2010) or intermittent mixing (30 min on–30 min off) (Persoone et al. 1980). The latter study investigated the impact of mixing on growth rates of Chlorella sp. cultured in ponds mixed by air-lift pumps and observed a 10 % decrease in growth rates when the culture was intermittently mixed compared to a continuously mixed culture.

Decreasing mixing velocity under 15 cm s−1 has previously been shown to have negative effects due to the promotion of excessive boundary layers (Richmond 2013) and low mixing velocities allowing cells to settle and accumulate in stagnant zones, which develop anaerobic conditions leading to propagation of anaerobic bacteria, thereby decreasing the cultivation efficiency (Hadiyanto et al. 2013). However, no studies have been done where paddle wheels were turned off completely during night time. In our work, paddle wheels were stopped for 10.5 and 9.5 h over the night-time period and maintained optimum mixing velocity (20 cm s−1) during the whole illuminated part of the day. In such scenario, all benefits from mixing are available for microalgae when photosynthesis occurs, that is to say, when thermal and oxygen gradients could affect the culture or when exposure to the solar irradiance is needed in order to ensure efficient nutrients uptake and growth.

Based on the modelling of Lundquist et al. (2010), who reported that a 17 % reduction of energy consumption could be achieved if paddle wheel-driven ponds (30-cm depth) are operated for 14 h at 25 cm s−1 paddle wheel velocity and slowed at night to 20 cm s−1 for 10 h, we estimate that overnight shutdown of paddle wheels could reduce power consumption up to 37 %. This is predicted to yield a 33 % reduction in power consumption-associated costs. This prediction is based on results reported by Rogers et al. (2014), who studied power consumption and associated costs on a continuously mixed paddle wheel-driven raceway pond system that maintained a culture velocity of 30 cm s−1 and an average productivity of 15 g m−2 day−1. Therefore, our results suggest a 37 % reduction of the total energy consumption for paddle wheel operation is possible with no significant impact on growth performance if mixing operations are stopped during night time in cultures under optimal conditions.

The ability to substantially cut power usage at night increases the ability to integrate outdoor open ponds with solar power and enabling such facilities to be ‘off-grid’, allowing the installation of algae farms in locations where electricity power would be very costly or even impossible to supply such as remote or undeveloped regions.

We have tested this procedure with a diatom, Amphora sp., with less favourable results (unpublished data). One possible explanation for the more favourable results reported here could be that T. suecica did not completely settle out of the water column due to their flagella. In fact, samples taken during the night and examined under microscope showed lively and fast moving cells. Additionally, this experiment was done in summer time, with a short night period. Longer nights would mean longer unmixed periods, and such situations need to be evaluated to reach a comprehensive understanding.

Irrespective of its design (open or closed), any large-scale algae cultivation facility needs to be reliable in the long term and stable; therefore, contamination issues are very relevant to a profitable commercial photobioreactor (Moheimani 2013). In the present experiment, contamination started to rise during a heat wave when the culture was stressed but contaminants were washed out once semi-continuous harvest could be resumed. Other studies suggest that high pH and salinity could be necessary conditions to prevent uncontrolled contamination of some algae species in open ponds (Wang et al. 2013). In this experiment, culture pH higher than 9.5 and to a less extent, salinity around 3.3–3.5 %, may have avoided protozoa/bacteria uncontrolled proliferation.

Efficient cultivation of microalgae requires maintenance of highest possible photochemical conversion efficiencies (Ralph and Gademann 2005), and therefore, photosynthetic activity was assessed during the 24-h follow-up. Measurements of Fv′/Fm′ provided information on the maximum efficiency of photon conversion to chemical energy at the PSII reaction center while exposed to actinic light (Cosgrove and Borowitzka 2006; Oxborough 2004). As for the diurnal pattern of effective quantum yield reported in Kurzbaum et al. (2010), Fv′/Fm′ decreased during the day and was highest at dawn and dusk. No significant difference in Fv′/Fm′ was observed between treatments at any time of the diurnal cycle. So, although mixing is an important operation for maintaining high productivity during the day, the absence of mixing overnight does not appear to impact photosynthetic performance as measured at PSII by PAM fluorometry during the day. However, oxygen and pH data indicate that inhibition of photosynthesis occurs during the middle of the day and much of the afternoon. This result is likely due to downstream processes such as photorespiration rather than photodamage at PSII and can be supported by the observed increased in pH after each harvesting, indicating a recovery of the carboxylase activity of Rubisco. These data add confidence to the conclusions made from comparing growth parameters of the three ponds. The combined indication is that T. suecica grown in outdoor paddle wheel-driven raceways is not impacted by being unmixed overnight and pre-mixing before sunrise is not required. This last aspect may be important for reducing any necessary reliance on batteries to store solar power.

Importantly, between-pond differences in culture performance were seen when unfavourable environmental conditions occurred. On the fourth day after the high temperature climatic event, which caused growth decline in all three ponds, a 1-week recovery phase started. A statistically significant variation in growth rate between T. suecica grown in continuously mixed conditions and the two non-mixed overnight ponds was observed. Average biomass productivity during this period was 42 % higher in the continuously mixed pond than in any of the non-mixed overnight cultures. This emphasizes the importance of rigorous and vigilant monitoring of environmental conditions and culture health when maintaining large-scale outdoor cultures and suggests that if there is any diminution in the growth rates, turning to 24-h mixing operation would aid to more effectively restore normal growth activity.

As far as the authors are aware, this is the first study investigating the effect of completely suspending night time mixing operations on microalgae growth. Although the scale attempted here is far from that of a commercial facility, the results are encouraging and suggest that mixing-associated power costs may be reduced by roughly a third. However, further scale-up is required before the concept is applied at a commercial scale. Importantly, the reduction of the powered operation of paddle wheels to daylight hours allows more effective integration with solar power, a key consideration, especially for remote or Third World regions.