Influence of culture conditions on the production of extracellular polymeric substances (EPS) by Arthrospira platensis

Arthrospira platensis is a cyanobacterium that is of great biotechnological interest, particularly for the food industry, as it possesses a high content of proteins, pigments, lipids and carbohydrates. Cyanobacteria produce extracellular polymeric substances (EPS), which are co-products of secondary metabolism that present thickening or gelling properties. A 3-level factorial design was used to study the combined effect of different nitrate concentrations and photon flux density (PFD) values to evaluate the biomass and EPS production of A. platensis. The best result in terms of biomass production was obtained under condition 6 (2 g L−1 NaNO3 and 600 μE m−2 s−1) yielding a concentration of 1.292 g L−1. However, condition 1 (0.25 g L−1 NaNO3 and 200 μE m−2 s−1) produced the greatest EPS yield (111 mg g−1), followed by condition 9 (2 g L−1 NaNO3 and 1000 μE m−2 s−1). FTIR analyses of EPS samples indicated the presence of carboxylate and sulfate functional groups, and rheological studies of the EPS at 5 and 10 g L−1 revealed a dilute solution behavior.


Introduction
The phylum of cyanobacteria comprises a group of diverse prokaryotes. These microorganisms typically use autotrophic metabolism, but can also be grown in heterotrophic and mixotrophic conditions, and are commonly found in brackish or marine environments (Baldev et al. 2015). They are responsible for the incorporation of a large proportion of atmospheric CO 2 and some of them are N 2 fixers (Oliver and Atsumi 2014). Cyanobacteria produce extracellular polymeric substances (EPS), which have a protective function in biofilm formation, a mechanical barrier against desiccation (Decho and Gutierrez 2017), and which also aids in the absorption of heavy metals (Goo et al. 2013). The composition of EPS depends on the species. Generally, they are formed by various substances, such as proteins, polysaccharides, lipids, humic-like substances, DNA, lipopolysaccharides and glycoprotein heteropolymers (Can et al. 2019). EPS may be entirely released into the extracellular environment or may be associated with the cell surface as sheaths, capsules or slime (Pereira et al. 2009).
EPS have attracted industrial interest because of their polyanionic character and that they are a renewable source. They can be used as thickeners, emulsifying agents and biosurfactants, with applications in the food and biomedical industry. Antibacterial (Mundt et al. 2001), antioxidant (Trabelsi et al. 2016) and anti-inflammatory properties (Xiao et al. 2018) have already been reported for EPS of cyanobacteria.
Arthrospira platensis, also called Spirulina platensis, is a filamentous blue-green cyanobacterium, belonging to the order Cyanophyceae, division Cyanophyta (Manirafasha et al. 2018). Its biomass is used in human food due to its high protein content (approximately 70% of dry weight). It also produces pigments such as carotenoids and phycocyanin, polyunsaturated fatty acids, several vitamins, minerals and other constituents with antioxidant activity (Gong et al. 2008;Shabana et al. 2017). Sulfated heteropolysaccharides, frequently formed by neutral (xylose, galactose, glucose, fructose, rhamnose, arabinose and mannose) and uronic acid (galacturonic and glucuronic acids) repeating units, have been identified in the EPS of this species (Trabelsi et al. 2009a). Literature reviews suggest that the EPS from A. platensis have specific properties, which may contribute to human health. A fraction containing sulfated polysaccharides, at concentrations between 18 and 36 μg mL −1 , was shown to inhibit the replication of Koi herpes virus (Reichert et al. 2017). Challouf et al. (2011) reported antibacterial activity against Salmonella typhimurium and Staphylococcus epidermis. Furthermore, antioxidant activity of crude EPS was described by Dejsungkranont et al. (2017).
Although EPS could contribute to numerous industrial areas, the cost of the current extraction and purification procedures, the low yield and the lack of knowledge regarding their structure impede their commercial use at this time. In order to minimize costs and enhance EPS yields, optimization of the culture parameters may be necessary. EPS production by cyanobacteria can be influenced by environmental and nutritional conditions, including temperature (Trabelsi et al. 2009a), photonic flux density (PFD) (Villay et al. 2013) and nitrogen starvation (Arad and Levy 2010). The right combination of these parameters could greatly improve EPS production. However, only a few studies have been conducted with the objective to enhance the EPS yield from species of cyanobacteria. For example, the temperature and PFD parameters were optimized for the production of EPS from Cyanothece sp., with 140 mg L −1 at 30 °C and 8 μE m −2 s −1 , and from Rhodella violacea, with 600 mg L −1 at 24 °C and 420 µE m −2 s −1 (Ohki et al. 2014;Villay et al. 2013). Whilst for Dunaliella salina and Synechocystis sp., the optimization of the required NaCl concentration was investigated, reaching EPS yields of 944 mg L −1 and 630 mg L −1 , respectively (Mishra and Jha 2009;Ozturk and Aslim 2010). Using A. platensis, factors of temperature, PFD and the concentrations of NaCl and NaNO 3 have been studied individually or in combination by Trabelsi et al. (2009b), Chentir et al. (2017) and Dejsungkranont et al. (2017), producing maximum yields of 210 mg L −1 , 0.98 g g −1 and 772.1 ± 192.9 mg g −1 , respectively.
The determination of the best techniques to extract and purify the EPS, as well as the full characterization of the resulting product, could also bring it closer to industrial use (Delattre et al. 2016). In general, the culture medium of the cyanobacteria is centrifuged or filtrated to remove the cells and then the supernatant or filtrate is concentrated to reduce the water content. Some authors have employed the use of a heat bath (Parikh and Madamwar 2006) or a process using a membrane, such as tangential ultrafiltration (Han et al. 2014) or microfiltration (Ahmed et al. 2014), to attain the concentrate. However, membrane processing depends on the viscosity of the culture medium, pore size distribution and transmembrane pressure . The most frequently used technique to recover the EPS from the concentrate is by alcohol precipitation. Under these conditions, some salts present in the culture medium may also co-precipitate. Ultrasoundassisted extraction (UAE) and microwave-assisted extraction (MAE) have also been used to extract EPS, although to a lesser extent (Budarin et al. 2012). It is worth noting that ultrasound waves may also degrade EPS macromolecules (Du et al. 2012).
As yet, no study in the literature has evaluated the combined effect of nitrate (NaNO 3 ) starvation and PFD on the EPS production from A. platensis culture. Nitrogen starvation is known as a stress condition, which favors the accumulation of some reserve products and the production of EPS from cyanobacteria (Lupi et al. 1994). An increase in PFD has also been related to an increase in EPS yields (Markou et al. 2012). However, the intracellular events that lead to EPS production are still being deciphered. Some authors have proposed two events in diatoms: (1) the combined effect of PFD and nitrate starvation reduce cellular growth and cause an excess of assimilated carbon, which is released as carbohydrates; (2) the electrons accumulated in the photosynthetic electron transport chain induce the production of reactive oxygen species (ROS) that can cause cellular damage, thus EPS are produced to act as a protection barrier against these ROS (Miklestad 1995;Piedras et al. 2010).
In this work, an experimental design methodology was used to investigate the combined effect of nitrogen concentration and PFD on the biomass concentration and EPS content from the cultivation of A. platensis. The resulting products were analyzed as for their protein, carbohydrate and metals composition. Furthermore, EPS samples were characterized by infrared spectroscopy, and for their thermogravimetric and dynamic rheological properties.

Microorganism
Arthrospira platensis was provided by the Elizabeth Aidar microalgae collection from the Fluminense Federal University (Niterói, RJ/Brazil). The cell suspension of A. platensis was first cultivated in 300 mL modified Zarrouk's medium at 32 °C, according to George (1976), to be used as an inoculum. The experiments were carried out at room temperature (23 °C) in Erlenmeyer flasks at a PFD of 150 μE m −2 s −1 and under constant stirring. All other reagents were purchased from Sigma Aldrich (São Paulo, SP, Brazil). Distilled and deionized water was used in all cases.

Culture media
Prior to experimental conditions, cultures were grown in 5-L bottles with 3700 mL of modified Zarrouk's medium (George 1976) and 300 mL of A. platensis inoculum, and incubated in a germination chamber at (32 ± 2) °C with constant aeration and white LED lighting at a PFD of 1000 μE m −2 s −1 . Measurements of the PFD were carried out with a QSL 2100 dosimeter from Biospherical Instruments Inc. (San Diego, CA, USA). As LED lamps have unequal values of PFD depending on the region of the bottle surface, through which the light passes, a complete mapping of PFD values was made along 36 points on the surface of the bottle.

Experimental design
Using the Design Expert program, version 10.0.6.0., a 3-level factorial design (3 2 ) was applied to plan culture conditions for the production of A. platensis biomass and EPS. A total of 12 runs were planned, nine of which for combinations of three levels and a central point in triplicate (Table 1). The independent variables were NaNO 3 concentration (0.5, 1.125, and 2 g L −1 ) and PFD (200, 600 and 1000 μE m −2 s −1 ). The two responses evaluated were biomass concentration (mg L −1 ) and EPS yield (mg g −1 ). The effect of each factor was determined by the analysis of variance (ANOVA) for a confidence level of 95%.

Monitoring of growth and determination of the final biomass concentration
A. platensis growth was monitored by daily measurements of the optical density (O.D.) of the cellular suspension at 730 nm for 21 days. The specific growth rate (μ) was determined over the exponential growth phase, and given by Eq. 1: where x 1 and x 2 are the biomass concentrations, and t 1 and t 2 are the times, corresponding to the beginning and the end of the exponential phase, respectively.
To determine the dry biomass weight, aliquots of 20 mL were taken from the culture medium at the end of each experiment and filtered through 0.7-1.2 μm Sartorius glass fiber membranes that had been previously weighed. After being extensively washed to eliminate excess of salt, the membranes were dried at 105 °C in an Ohaus ® model MB45 Moisture Analyzer balance for approximately 15 min, and weighed, giving the dry biomass (g) by difference.

EPS production
To recover the EPS from the extracellular medium, the procedure reported by Parikh and Madamwar (2006) was followed. Briefly, the filtered culture medium, which still contained some biomass, was centrifuged at 13,000 rpm for 20 min in a Hettich Rotixa centrifuge, model 420R (Andreas Hettich GmbH & Co. KG, Tuttlingen, Germany). The resulting supernatant was heated at 70 °C for 12 h, reducing its volume to ¼. The EPS were then recovered by adding ice-cooled ethyl alcohol. After 12 h at 4 °C, the product was filtered and completely dried in an oven at 60 °C. To eliminate residual salt, the samples were resolubilized in distilled water and then dialyzed against water under constant stirring at room temperature for 24 h. The final product was recovered by adding 95% ethyl alcohol and lyophilized.

Chemical and physicochemical characterization of EPS
The methodology proposed by Dubois et al. (1956) was used to quantify the total carbohydrate content for each recovered EPS sample, by adding H 2 SO 4 concentrated at 80% (m/v) and phenol at 5% (v/v). A standard curve was obtained using d-glucose. Soluble protein was determined by the Folin's method, using bovine serum albumin as the standard (Lowry et al. 1951).
Fourier transform infrared spectroscopy (FTIR) analyses were carried out using a Perkin Elmer spectrometer, Frontier model (Waltham, MA, USA) at room temperature, using KBr disks, in the 4000-400 cm −1 range, with accumulation of 20 scans and 4 cm −1 resolution.
Thermogravimetric analyses were carried out for EPS samples under nitrogen atmosphere on a TGA Q-500 equipment from TA Instruments (New Castle, DE, USA). Approximately, 10 mg of sample were heated from 20 to 700 °C, at a 10 °C/min rate.
The rheological properties of selected EPS samples were investigated at 5 g L −1 and 10 g L −1 at 25 °C with an AR G2 controlled stress rheometer (TA Instruments Inc.), equipped with a coaxial cylinder geometry. Initially, strain sweep tests were conducted as the variation of complex modulus at 6.28 rad/s to determine the linear viscoelastic range. After a period of 10 min, an oscillatory frequency scan was performed, from 10 −1 to 7 × 10 2 rad s −1 (with a deformation value of 10%), within the region of linear viscoelasticity. Finally, viscosity changes under steady-flow regime were investigated as a function of shear rate, from 10 to 10 2 s −1 .

Statistical analysis
The carbohydrate and protein quantifications were carried out in triplicate and the results were expressed as mean values ± standard deviation (SD). One-way ANOVA was applied to the data by using the PAST 3.20 software, available at https ://folk.uio.no/ohamm er/past. The mean values were compared by the Tukey's test considering a confidence level of 95% and level of significance (p < 0.05).

Evaluation of Arthrospira platensis growth under different cultivation conditions
The growth curves of A. platensis cultivated in different culture conditions are presented in Fig. 1. As observed, none of the conditions presented an adaptation phase. Condition 6 resulted in the highest growth rate (µ = 0.55), in medium with 2 g L −1 of NaNO 3 and under 600 μE m −2 s −1 of PFD. Condition 2 led to the lowest growth rate, with 1.125 g L −1 of NaNO 3 and a PFD of 200 μE m −2 s −1 , followed closely by number 1 (0.25 g L −1 of NaNO 3 and 200 μE m −2 s −1 of PFD) and condition 3 (2 g L −1 of NaNO 3 and 200 μE m −2 s −1 of PFD). All three of these conditions corresponded to the lowest PFD used. The results demonstrate that the A. platensis growth rate was practically independent of the NaNO 3 concentration, and indicates that the PFD was the limiting factor.
The PFD values may have affected the growth rate, which increased until reaching a point of light saturation and, from this point, there may even have been a decrease in the rate of growth as a consequence of photo inhibition (Carvalho et al. 2011). In cultures grown at 1000 μE m −2 s −1 and with NaNO 3 at a concentration of 2 g L −1 , light saturation may have occurred, which resulted in a lower rate of cell growth. In cultures using a PFD at 600 μE m −2 s −1 , the NaNO 3 concentration was probably the limiting factor for A. platensis growth.

Final biomass concentration and EPS yield under different growth conditions
Some researchers reported that different concentrations of NaNO 3 and PFD could induce the EPS production from cyanobacteria (Aikawa et al. 2012;Dejsungkranont et al. 2017;Ohki et al. 2014;Villay et al. 2013). Modified Zarrouk's medium has 2.5 g L −1 of NaNO 3 . Therefore, for the experimental design, a NaNO 3 concentration (2 g L −1 ) close to that for which regular cell growth is already known and a concentration reduced by 75% (0.5 g L −1 ) were used. For the PFD, the minimum and maximum values that have reportedly been used for A. The values of the final biomass concentration and the EPS yield are given in Table 1. In general, the O.D. values were related to the biomass concentration. However, there was some variation for the intermediate values, although the differences between the expected O.D. and the observed O.D. were not so high, suggesting they could have been caused by small measurements error.
Considering the results, the highest final biomass concentration (1.292 g L −1 ) was observed under conditions with the highest NaNO 3 concentration (2 g L −1 ). However, for the EPS yield, the best conditions were achieved under the lowest NaNO 3 (0.25 g L −1 ) and PFD (200 μE m −2 s −1 ), which generated 111 mg g −1 of EPS. The second highest EPS content (100 mg g −1 ) was obtained under conditions of the highest values of NaNO 3 (2 g L −1 ) and PFD (1000 μE m −2 s −1 ). In other words, the maximum EPS content (EPS mass/biomass) was attained using both extremes of the experimental design.
The results presented in Fig. 2a, b show that the EPS production by A. platensis was not directly associated to its growth. As can be observed, the response in which the highest final biomass concentration was obtained differed from that which generated the highest EPS production. In general, metabolic stress may negatively affect cyanobacteria growth, during which cells use their energy to produce reserve compounds such as carbohydrates and EPS (Santos et al. 2019). Condition 9 was an exception, since this condition promoted one of the highest EPS yields and, at the same time, one of the highest final biomass concentrations. Table 2 shows the results of the coefficients and their interactions, R 2 , lack of fit and the p-values for final biomass concentration and EPS yield from A. platensis. A 95% confidence level was adopted (p < 0.05). The model was significant for final biomass concentration with a p value equal to 0.0379. It was found that the PFD and its interactions have a significant influence on the model. For the EPS production, the model was significant, with a p-value equal to 0.0164, and the variables that showed a significant influence were the PFD and its interactions.
The regression equation was generated, as shown in Eqs. 2, 3, according to the values presented in Table 2: (2) Biomass concentration = 1.07 + 0.0755A + 0.0892B It is noteworthy that the lack of fit values for both biomass and EPS production were significant (p < 0.05). In this case, the model does not have a good predictive capacity. Despite the lack of adjustment, it is possible to visualize the best working regions.
In Fig. 2a, the response surface graphics show that the conditions which led to the highest final biomass concentration were a PFD of 600 μE m −2 s −1 and 2 g L −1 NaNO 3 . A decrease in biomass production was also observed when the value of PFD was smaller. The PFD can influence cyanobacteria photosynthesis. Excessive or insufficient light may affect the biomass productivity and yield of metabolic products (Carvalho et al. 2011). It is likely that the nitrogen-reduced conditions in this work did not compromise the biomass production and were high enough to sustain the cell growth. However, the lowest PFD negatively affected the biomass production. In this case, it seems that light intensity was insufficient to support the biomass production. Aikawa et al. (2012) also observed that biomass production was related to the PFD, and that A. platensis produced the highest final biomass concentration (1.6 g L −1 ) at 700 μE m −2 s −1 , the highest PFD value they tested. Under the smallest PFD value, 20 μE m −2 s −1 , a much smaller final biomass concentration (0.1 g L −1 ) was attained. Figure 2b shows the response surface obtained for the EPS content (mg g −1 ). The highest values were observed at the two extremes, 111 and 100 mg g −1 , using conditions 1 and 9, respectively, in addition to condition 3, which also generated an EPS yield of 100 mg g −1 . The EPS production from A. platensis was also affected by the PFD; however, it was not the only determinant variable. NaNO 3 starvation contributes to the increase in the C/N ratio. Consequently, it promotes the incorporation of carbon reserve in the EPS (Otero and Vincenzini 2003). The lowest PFD (200 μE m −2 s −1 ) and NaNO 3 concentration (0.5 g L −1 ) resulted in the highest EPS yield. It is possible that the combined effects led to cells producing more EPS in response to the limited conditions. This result is in accordance with that found by Chentir et al. (2018), who obtained the highest EPS yield, 0.902 g g −1 , using 0.5 g L −1 NaNO 3 combined with a PFD of 57.2 μE m −2 s −1 in A. platensis. Chentir et al. (2017) evaluated the maximization of EPS production from A. platensis as a function of variations in NaCl concentration and PFD. Although the PFD did not have any positive effect on EPS production, its interaction with the NaCl concentration provided a 0.98 g g −1 yield of EPS. On the other hand, Dejsungkranont et al. (2017), through studying the effect of PFD on the production of EPS from A. platensis, observed that the highest level of PFD (203 μE m −2 s −1 ) favored EPS production  Trabelsi et al. (2009a) carried out a study evaluating the effect of different temperatures and PFD on the final biomass and EPS concentrations by A. platensis, and described a possible correlation between the two responses. According to the authors, the production of biomass and EPS are mutually dependent, and the increase of EPS production may be associated with the kinetics of growth. To achieve a greater EPS content, it is necessary to optimize the PFD while the temperature should be maintained between 30 and 35 °C. In that work, A. platensis produced the maximum EPS content at the highest PFD used, 180 μE m −2 s −1 , with 297.4 ± 11.1 mg L −1 . The result for EPS content found by Trabelsi et al. (2009a), 210 mg L −1 , under 100 μE m −2 s −1 and 2.5 g L −1 NaNO 3 , was twice that of our result, 91 mg L −1 .
The results found in the present study demonstrate that the choice of the best culture conditions depended on the response of interest. The results in Fig. 2a, b showed that the production of EPS by A. platensis was not directly associated to growth. Nevertheless, it is possible to choose a condition in which the EPS content as well as biomass concentration could be produced at reasonable values.
Other studies in the literature using different microorganisms have reported the effect of culture conditions on the EPS content and biomass concentration. For Cyanothece sp. 113, NaNO 3 concentrations between 0 and 200 mg L −1 were used (Su et al. 2007). In this case, the final biomass concentration was observed to increase, reaching 1.2 g L −1 with a NaNO 3 concentration of 74.3 mg L −1 , but then decreased at values higher than 100 mg L −1 . However, a decrease in EPS concentration, from 7 g L −1 to 5 g L −1 , was reported elsewhere when 200 mg L −1 of NaNO 3 was used. In the same work, the PFD effect was evaluated in the 20-100 μE m −2 s −1 range and the best condition was found to be 86 μE m −2 s −1 for both biomass and EPS production (Su et al. 2007). Figure 3 shows the FTIR spectrum for EPS samples from A. platensis. The broad bands observed around 3400 cm −1 are attributed to O-H and N-H stretching. The weak absorptions in the region 3000 cm −1 to 2840 cm −1 are associated to C-H asymmetrical and symmetrical stretching modes of methyl and methylene groups. The absorption at 1650 cm −1 is attributed to C = O stretching of carboxylate and amide groups (amide I band). The high-intensity band with maximum at 1442 cm −1 is attributed to more complex vibrations, associated to O-H bending (Can et al. 2019;Trabelsi et al. 2009a). The most intense absorption of the spectrum, at 1046 cm −1 , may be attributed to C-O-C, S-O and P-O-C stretching vibrations. This result evidenced the presence of polysaccharides, proteins/polypeptides, and of sulfate and phosphate groups linked to polymeric substances.

Thermal analysis
Thermogravimetric analysis is an important technique, as it demonstrates the thermal stability of the EPS (Fig. 4). The EPS displayed two stages of thermal degradation. In the first stage, up to 150 °C, 10% weight was lost and this may be attributed to the loss of water and other volatile substances. In the second stage, the degradation of the polymer chain occurred, between 225 and 350 °C, with 50% of the total mass lost. The thermal degradation of EPS from microalgae and cyanobacterium was recently reported. Two stages were observed for the EPS from Dunaliella salina; in the first, 15% weight was lost at up to 150 °C. In the second, 55% of total EPS weight loss was observed with maximum loss at 240 °C (Mishra et al. 2011). On the other hand, for the EPS from Nostoc carneum, three stages of thermal degradation were detected. In the first stage, 15% weight was lost up to 155 °C. In the second stage, with a maximum at 237 °C, 39% weight was lost, attributed to polysaccharide degradation. The third phase occurred up to 378 °C with 32% of weight loss (Hussein et al. 2015).

Carbohydrate and protein quantification
The chemical composition of the EPS depends on environmental conditions and the microorganism studied (Nouha et al. 2018). According to Wingender et al. (1999), the majority of EPS constituents are carbohydrates and proteins. In A. platensis, the carbohydrate and protein content were estimated as 55% and 13%, respectively, in a photoautotrophic growth for 25 days (Pignolet et al. 2013). Carbohydrate and protein contents for A. platensis EPS determined in the present work are shown in Table 3. The highest contents of carbohydrates, 39.5 ± 2.34 mg g −1 and 38.73 ± 2.55 mg g −1 , were observed under conditions 5 (1.125 gL −1 NaNO 3 and 600 μE m −2 s −1 ) and 9 (2 gL −1 NaNO 3 and 1000 μE m −2 s −1 ), respectively.

Rheological properties
Aqueous solutions of EPS 01 (0.25 gL −1 NaNO 3 and 200 μE m −2 s −1 ) and EPS 09 (2 gL −1 NaNO 3 and 1000 μE m −2 s −1 ) were prepared at 5 g L −1 and 10 g L −1 , and their rheological properties were investigated. Figure 5a, b shows the variation of storage modulus, G′, and loss modulus, G′′, with the oscillatory frequency. Their rheological behaviors were quite similar to that previously observed for other polysaccharides, such as gum algaroba, a galactomannan extracted from Prosopis juliflora (Azero and Andrade 2006). At very low frequencies, the viscous character predominated. After the crossing of G′ with G′′, the elastic character was predominant. As expected, the value of the frequency at which the crossing occurs was smaller for the highest concentration. The same behavior was observed for the EPS09 supernatant at 5 and 10 g L −1 .
Our results differ from those reported by Chentir et al. (2017), who also assessed the EPS of A. platensis. The authors described a gel-like behavior, in which the  Table 3 Carbohydrates and proteins in different culture conditions Statistical differences (One-way ANOVA followed by Tukey's test; p < 0.05) among conditions are represented as different letters

Conclusions
The present work demonstrated that combined values of NaNO 3 and PFD were capable of enhancing the EPS yield and biomass concentration of Arthrospira platensis. The EPS production was not associated with the growth rate, and under nitrate starvation and the lowest PFD, the highest EPS production was achieved. The FTIR spectra showed that the EPS contained sulfate groups, carbohydrate and proteins, while the rheological studies suggested that the EPS exhibited a dilute solution behavior.
In terms of further enhancing the biomass concentration and EPS yield of A. platensis, whilst also reducing production costs, alternative culture media could next be evaluated.