Monitoring indigenous microalgae using derivative spectrophotometry and comparison with M. aeruginosa and C. vulgaris

BOD

Biochemical Oxygen Demand

COD

Chemical Oxygen Demand

DOC

Dissolved Organic Carbon

MDL

Method Detection Limit

UV-Vis

Ultraviolet-visible

BG-11

Blue-Green Medium

BBM

Bold's Basal Medium

3N-BBM

Bold's Basal Medium with 3-fold Nitrogen and Vitamins

Microalgae are a diverse group of organisms that thrive in the presence of nutrients and under warm temperatures. The relative ease with which microalgae grow can be both advantageous and disadvantageous. Microalgae can be commercially cultivated and harvested and used as a source for bioenergy generation (Menetrez 2012; Scaife et al. 2015; Laurens 2017; Laurens et al. 2017) or water treatment (Jaroo et al. 2019; Moayedi et al. 2019; Kerich 2020), but undesired growth of algal blooms in water bodies can lead to environmental problems (Winter et al. 2011).

Several microalgae species have demonstrated remarkable pharmacological and biological qualities (Khan et al. 2018). One such species is C. vulgaris – which is rich in micro and macronutrients including protein, iron, and vitamin B12 and is sold as a health supplement (Safi et al. 2014). C. vulgaris is also widely present in surface waters around the world. On the other hand, there are other species of microalgae that can have adverse impacts on human health and the environment. For example, cyanobacteria, commonly called blue-green algae, are prokaryotic organisms that flourish in ponds, lakes, wetlands, and other surface waters. Cyanobacteria algal blooms need to be consistently checked during the summer months as they can pose a serious threat to water quality, ecosystem stability, and public health through toxin production (Leland et al. 2019). Due to the rapid increase in cyanobacteria blooms caused by water pollution and climate change, they have become the recent focus of scientific research to develop real-time analytical tools for early detection and monitoring.

Due to an increasing demand to develop real-time analytical tools, several devices and methods have been employed to detect and monitor microalgae and cyanobacteria in waterbodies. Traditionally, microscopy, hemocytometry, and UV-Vis spectroscopy (Picazo et al. 2013) have been used to evaluate both algal cultures and environmental samples. Over the years, automated nucleic acid biosensors have been developed for in-situ monitoring (Metfies et al. 2009), and methods such as chromatography (Picazo et al. 2013), spectrofluorometry (Gregor & Maršálek 2004; Bowling et al. 2016; Gsponer et al. 2018), and flow cytometry (Cunningham 2002) have also been used for detailed analyses that examine the ability of microalgae to grow and form algal blooms based on their chlorophyll properties.

Recent studies by Agberien et al. (Agberien & Örmeci 2019) and Almomani et al. (Almomani & Örmeci 2018) looked into the detection and monitoring of cyanobacteria and C. vulgaris in deionized and surface waters using derivative spectrophotometry and Savitzky-Golay algorithm. The studies showed that the use of derivative spectrophotometry along with mathematical tools found an improvement in the detection limit of the microalgae samples studied.

This study aims to investigate the monitoring of mixed indigenous microalgae with derivative spectrophotometry and Savitzky-Golay algorithm as a possible real-time monitoring tool for surface waters and assess the spectral fingerprints of an indigenous microalgae culture in comparison with pure cultures of Chlorella and cyanobacteria strains to improve the specificity and selectivity of detection for different microalgae strains and mixed cultures.

The samples of indigenous microalgae were collected from Dow's Lake in Ottawa, Canada. During the summer months, due to the slow speed of water and sunlight combined with warm temperatures, the lake is overwhelmed with algal blooms and is often deemed unsafe for public use (Confirmed: Blue-green algae responsible for strange rideau canal tint; Wang et al. 2017). The algal samples were obtained by following the standard procedures for safe collecting and handling of water samples that may contain toxins (Queensland Government 2018).

Pure cultures of M. aeruginosa CPCC 632 and C. vulgaris CPCC 99 along with sterile growth media BG-11 (Stanier et al. 1971), BBM and 3N-BBM were purchased from the Canadian Phycological Culture Centre (CPCC) at the University of Waterloo, Ontario, Canada.

Due to uncertainty on the growth media for the cultivation of indigenous microalgae samples, BG-11, BBM and 3N-BBM were initially tested to determine which one would work best for the indigenous microalgae sample. Each sterile growth medium was inoculated with 1 mL of the indigenous culture sample such that the dilution ratio was 1:100. To prevent contamination, the 250 mL Erlenmeyer flasks used to culture the indigenous culture sample were rinsed with deionised water, after which the mouth of each flask was covered in aluminum foil, autoclaved at 15 psi and 121 °C for 30 minutes, and cooled at room temperature. The cultures were incubated in a temperature-controlled incubator and maintained at 25 °C under 24-hour light intensity of 1,800–2,500 lux. Each flask was manually stirred twice a day at 8-hour time intervals or more. It was noted that BBM was the best performing growth medium, with samples reaching ample algal population growth within 12 days, whereas BG-11 was the worst growing medium for the indigenous microalgae culture sample as the populations did not thrive in this medium. 3N-BBM showed a delay in growth with requiring sub-culturing once every 19 days. Subsequently, the indigenous microalgae sample in sterile BBM was sub-cultured every 12 days until the study was completed.

In order to prepare the samples for cell counting, the indigenous microalgae, M. aeruginosa and C. vulgaris samples were required to be centrifuged to concentrate the microalgae. Several cycles of centrifuge at 8,000 g for 5 minutes, each consisting of 40 mL of the sample, were required to obtain an adequate concentration of microalgae. Once the samples were ready, cell counting was carried out using a light microscope and Neubauer chamber (hemocytometry). All three samples were separately viewed under the microscope, and the number of cells was counted using a Neubauer chamber.

After enumerating the various cultures under the light microscope and counting the number of cells per mL, dilutions of the samples were prepared using deionised water. The same dilution ratios and concentration ranges were used to enable comparisons of results of indigenous microalgae, M. aeruginosa and C. vulgaris samples. After dilution, concentrations of samples ranged from 23,675 cells/mL to 12,121,856 cells/mL for indigenous microalgae sample, 42,266 cells/mL to 29,680,000 cells/mL for when indigenous microalgae sample was spiked with M. aeruginosa and lastly 42,266 cells/mL to 21,640,118 cells/mL when indigenous microalgae sample was spiked with C. vulgaris. These concentration ranges were selected based on the low, medium, and high microalgae concentrations in WHO guidelines.

Spectrophotometric analysis was done using a Cary 100 Bio UV-Vis Spectrophotometer (Agilent Technologies) (Cary 100/300/4000/5000/6000i/7000 Spectrophotometers User‘s Guide) and a 5 cm quartz cuvette. An absorbance scan was done between 200 to 800 nm; however, the analysis focused on the absorption of light between 400 to 800 nm due to the presence of important peaks in this range. The spectrophotometer was set to baseline subtraction to account for absorption by blanks (water samples), after which absorption spectra for respective growth media were taken and subtracted from final absorbance measurements (growth media and respective microalgae sample). Standard calibration curves were plotted at low, medium, and high concentration ranges to validate whether spectral analysis agreed with the Beer-Lambert law.

The first-order derivative of absorbance refers to the rate of change in absorbance with respect to wavelength (∂A/∂λ). The subsequent orders of derivatives were also calculated using the aforementioned definition. The various graphs were plotted using changes in absorbance values at 1 nm bandwidth.

Savitzky-Golay coefficients for the first derivative of absorbance were used to smoothen the plot of the first derivative (Savitzky & Golay 1964; Kus et al. 1996).

Twenty-three data points were used for smoothing such that m = 23; i=−11, −10, ….10, 11; N = 1,012; Ci were fitted according to Savitzky-Golay coefficients for the first derivative of absorbance.

The Method Detection Limit (MDL) was calculated according to the Hubaux and Vos method (Hubaux & Vos 1970). In this method, three replicates for every concentration range were selected, and the MDL tool developed by Chemiasoft was used to carry out these calculations.

Several different algal species were present in the indigenous microalgae sample, and Scenedesmus, cyanobacteria, Chlorella, and Oscillatoria along with diatoms, protozoa, and bacteria were identified under the light microscope. Figure 1 illustrates some of the microalgae and other microorganisms that were observed under the microscope.

The absorbance spectra for indigenous culture (Figure 2) for three concentration ranges (high, medium, low concentrations as per the WHO guidelines Chorus & Bartram 1999) exhibited peaks at approximately 445 nm, 640 nm, and 684 nm, with the most significant peak located at 684 nm. There was a strong linear relationship between absorbance and indigenous microalgae concentration with R² > 0.95 (Supplementary material, Figure S1). The method detection limit for absorbance at 684 nm in deionised water was calculated to be 158,693 cells/mL.

In order to gain a better understanding of the data, the first derivative of absorbance against wavelength was calculated and plotted in Figure 3. The first derivative of absorbance which is defined as the rate of change of absorbance of the microalgal sample in comparison to zero-order derivative which corresponds to absorbance of the microalgal sample. First-order derivative spectra showed that the signals were readily detected for high and medium concentrations of indigenous microalgae but differentiating between signal and noise became increasingly difficult for low concentrations. This increase in amplification of noise in low concentrations is due to the application of derivative spectrophotometry and it was also noted that the absorbance peak of 684 nm has shifted to 694 nm due to the application of derivative spectrophotometry.

The second derivative was calculated and plotted to determine whether the peaks could be distinguished more clearly. But this was not the case as the number of peaks doubled compared to that of the first derivative, and the difference between signal and noise was not distinguishable (Figure S2, Supplementary material). Hence, the second derivative for all cases was not considered further.

Since the second derivative of absorbance was not successful, an alternative method was followed where Savitzky-Golay first derivative of absorbance for high, medium, and low concentrations was calculated and plotted in Figure 4. The low concentration signals could now be detected, which was not possible in the first derivative of absorbance. Again, there was a strong linear relationship between the absorbance values and microalgae concentrations with R² > 0.95 (Supplementary material, Figure S3), and the method detection limit was successfully lowered to 60,170 cells/mL.

The results from the indigenous microalgae culture are summarized in Table 1, which shows the concentration ranges, peak absorbance wavelengths, R² values and the detection limits.

In this experiment, the indigenous culture sample was spiked with an equal concentration of M. aeruginosa and a similar process as above was carried out to observe the changes in the absorbance spectra and fingerprint of the indigenous microalgae culture when pure strain M. aeruginosa was added to it. The absorbance values of the sample increased significantly due to the concentration increase, while the number of significant peaks decreased from three to two (Figure 5) as compared to the absorbance spectra of the indigenous microalgae (Figure 2). In the mixed indigenous microalgae with M. aeruginosa, the peak at 640 nm was diminished, and the chlorophyll-a peak at 681 nm was steeper. The method detection limit was calculated to be 352,410 cells/mL for the zero-order derivative measurements (Table 2), which was poorer than the method detection limit of the indigenous microalgae culture at 158,639 (Table 1). Similarly, the first derivative of absorbance was calculated and plotted in Figure 6. The 681 nm absorbance peak was again more distinct and noticeable compared to the indigenous culture and shifted to 691 nm due to the application of derivative spectrophotometry.

Due to the inability of the first derivative of absorbance spectra to provide a clear difference between the signal and noise and the advantages of Savitzky-Golay transformation as seen for the previous case, Savitzky-Golay first derivative of absorbance was calculated and plotted in Figure 7. The smoothening of the spectra resulted in clear and distinguishable peaks, particularly in the low concentration range, and also satisfied the Beer-Lambert law (R² > 0.96). The method detection limit was calculated as 150,079 cells/mL, which is more than a two-fold improvement in detection compared to the zero-order derivative measurements. However, the detection limit was significantly worse compared to the indigenous microalgae culture (60,170 cells/mL). The results from the indigenous microalgae culture mixed with M. aeruginosa are summarized in Table 2, which shows the concentration ranges, peak absorbance wavelengths, R² values and the detection limits.

The indigenous microalgae culture was mixed with C. vulgaris in equal concentration and similar procedures were followed as shown for the previous two scenarios. Figures 8–10 show the absorbance, first derivative of absorbance, and the Savitzky-Golay first derivative of absorbance, respectively. The Savitzky-Golay first derivative of absorbance yielded the best results again with a method detection limit of 181,246 cells/ml compared to the method detection limit of the zero-order derivative (280,198 cells/mL). The results for the indigenous microalgae culture mixed with M. aeruginosa are summarized in Table 3.

The absorbance spectra of standalone indigenous microalgae culture and the indigenous microalgae culture mixed with C. vulgaris and M. aeruginosa were analyzed individually and their absorbance values at a selected low concentration (approximately 45,000 cells/mL) were compared against each other along with the absorbance values of pure strains of C. vulgaris and M. aeruginosa (Figure 11). At a similar concentration, M. aeruginosa had the highest absorbance, followed by C. vulgaris and the mixed microalgae sample, and they also had different absorbance fingerprints. For example, the chlorophyll-a peak at around 680 nm was present for all samples, but the chlorophyll-b peak at around 640 nm was visible for the indigenous microalgae culture. Samples that had indigenous microalgae mixed with C. vulgaris or M. aeruginosa took the characteristics of both samples and also had the average absorbance of both. The Savitzky-Golay first derivative of absorbance resulted in distinctive peaks and spectra fingerprints compared to the zero-order absorbance spectra indicating that they can potentially be used not only to detect but also to identify various species of microalgae in water bodies (Figure 12).

It is known from previous research that the peaks around 680 nm wavelength are indicative of chlorophyll-a whereas the peak around 640 nm shows the presence of chlorophyll-b (Frank 1946; Brown 1969). The peak around 445 nm shows characteristics of both chlorophyll-a and chlorophyll-b as the former pigment peaks around 400–450 nm and the latter peaks between 425 and 475 nm. The results presented in this study show that there is an increase in absorbance values at these wavelengths with increasing microalgae concentrations. At low microalgae concentrations, the peaks at the above-mentioned wavelengths fade, and noise starts to take over, thus making it difficult to distinguish between the peaks and noise.

The first derivative reveals twice the number of peaks given by the zero-order absorbance plot (Kus et al. 1996). This can be attributed to the very definition of the first derivative of absorbance, which is defined as the rate of change of absorbance of the microalgal sample in contrast to the zero-order derivative, which corresponds to the absorbance of the microalgal sample. As explained by Kus et al. (1996), a first-order derivative passes through zero at the wavelength where the maximum absorbance value is observed. This, in turn, shifts the peak wavelength of absorbance spectra when the first-order of derivative is plotted. However, a limitation of the derivatization process is that the signal-to-noise ratio decreases with derivative order. This unwanted effect is due to rapid, random changes of noise amplitude in the spectrum (Lettre et al. 2010) and utilization of a digital filter such as the Savitzky-Golay polynomial smoothening algorithm can circumvent this limitation as shown in this study.

The results presented in this study show that derivative spectrophotometry could be adopted as a detection tool for real-time monitoring of indigenous microalgae. The detection limit of spectrophotometry alone is not very sensitive, but through this research, it was found that the detection limit can be greatly improved by employing mathematical and statistical tools such as the Savitzky-Golay first derivative of absorbance. The detection limit for the indigenous microalgae sample was calculated to be approximately 158,693 cells/mL using absorbance measurements. But using Savitzky-Golay first derivative of absorbance, the detection limit was reduced to approximately 60,170 cells/mL, which is a more than 2-fold reduction in detection limit. Similar reductions in detection limits were found when the indigenous microalgae sample was mixed with M. aeruginosa and C. vulgaris in equal concentrations. Mixing the microalgae sample with M. aeruginosa and C. vulgaris also helped to understand how their presence affected the absorbance fingerprints of the indigenous microalgae sample and that they could still be detected by employing the Savitzky-Golay algorithm. Although the Savitzky-Golay algorithm sufficiently improves the method detection limit of microalgae, it was still not able to improve the detection limit to a range of low probability of adverse health effects (<20,000 cells/mL) established by the WHO guidelines (Kus et al. 1996).

The main advantages of using spectrophotometry are its simplicity, real-time capability, and ability to measure a wide range of water quality parameters (e.g., UV254, DOC, BOD, COD, turbidity, nitrate, and nitrite) simultaneously together with the cyanobacteria concentrations. No sample processing, treatment with reagents, or pigment extraction are required for the method and there is a well-established market (e.g., RealTech UV-VIS sensors) for real-time spectrophotometers that can scan the complete visible and ultraviolet range and analyze results in real-time.

The main disadvantage of spectrophotometry is that it neither has the low detection limits nor the specificity of fluorometry. Fluorometric scans have both excitation and emission spectra, and each compound has its unique fluorometric spectra, commonly referred to as its fluorescence signature. And hence, fluorometry is a more applicable tool with higher sensitivity in the monitoring of microalgae compared to spectrophotometry.

For future developments on this study, the detection limit can further be improved using different pathlengths for absorbance measurements, employing signal processing and new mathematical tools should be investigated.

As noted from various studies (Winter et al. 2011; Menetrez 2012; Scaife et al. 2015; Laurens 2017; Laurens et al. 2017; Jaroo et al. 2019; Moayedi et al. 2019; Kerich 2020), different forms of microalgae are being cultivated to be used to carry out water and wastewater treatment, desalination and bioenergy generation; but it is equally important to detect the levels of microalgae in our water bodies in order to avoid harmful microalgal blooms causing adverse health effects to humans and other organisms in the environment. Hence, this study looked into exploring derivative spectrophotometry as a tool to detect and monitor indigenous microalgae concentrations in water. The results of this study and comparing it with previous work by Agberien et al. (Agberien & Örmeci 2019) and Almomani et al. (Almomani & Örmeci 2018) show that Savitzky-Golay first derivative of absorbance performs better than the zero-order absorbance and first derivative of absorbance for detecting microalgae. The method detection limit with the Savitzky-Golay first derivative of absorbance was found to be 60,170 cells/mL for the indigenous microalgae as opposed to 158,693 cells/mL with the zero-order absorbance. This improvement in detection implies that indigenous microalgae concentrations could be monitored at relatively early stages using Savitzky-Golay derivative spectrophotometry. The method can also be successfully employed in commercial and industrial applications where real-time monitoring and quantification of microalgae are required. Furthermore, Savitzky-Golay derivative spectrophotometry can also provide fingerprints to increase the specificity and selectivity of detection of different microalgae strains and mixed cultures. The proposed method would benefit from further improvements to bring the detection limit below 20,000 cells/mL as per the WHO guidelines, which can be achieved by employing additional digital filters or changing the pathlength of absorbance measurements.

This research was funded by the Natural Sciences and Engineering Research Council of Canada (NSERC) under the Discovery Grants program.

All relevant data are included in the paper or its Supplementary Information.