By Andy May
We have been told that the phytoplankton population is declining rapidly around the world and, of course, the cause is climate change. Phytoplankton is the base of the ocean food chain and it accounts for about half of global primary productivity or organic matter creation (Boyce, Lewis and Worm 2010). Phytoplankton is the major consumer of carbon dioxide, the dreaded demon trace gas, and the major producer of oxygen. So, first question, is the estimated decline in phytoplankton accurate, significant or unusual? Second question, if the decline is real, are the measurements long term enough to show it is not a natural occurrence? What is the natural variability and how do we know man-made climate change is to blame? Let’s investigate this.
The purported problem
An article in Nature (Boyce, Lewis and Worm 2010) entitled “Global phytoplankton decline over the past century” was widely read and discussed at the time. The following is from the paper:
“We conclude that global phytoplankton concentration has declined over the past century; this decline will need to be considered in future studies of marine ecosystems, geochemical cycling, ocean circulation and fisheries.”
They conclude that the decline is due to higher sea surface temperatures (SST). This is somewhat counter-intuitive since higher temperatures usually correlate with more plant growth, not less. But, they state that their conclusions and the decline are “unequivocal.”
Over the following year numerous critical replies were also published in Nature, one had the title “Is there a decline in marine phytoplankton?” (McQuatters-Gollop, et al. 2011). The following is from this reply:
“Boyce et al. compiled a chlorophyll index by combining in situ chlorophyll and Secchi disk depth measurements that spanned a more than 100-year time period and showed a decrease in marine phytoplankton biomass of approximately 1% of the global median per year over the past century. Eight decades of data on phytoplankton biomass collected in the North Atlantic by the Continuous Plankton Recorder (CPR) survey, however, show an increase in an index of chlorophyll (Phytoplankton Colour Index) in both the Northeast and Northwest Atlantic basins …, and other long-term time series, including the Hawaii Ocean Time-series (HOT), the Bermuda Atlantic Time Series (BATS) and the California Cooperative Oceanic Fisheries Investigations (CalCOFI) also indicate increased phytoplankton biomass over the last 20–50 years. These findings, which were not discussed by Boyce et al., are not in accordance with their conclusions and illustrate the importance of using consistent observations when estimating long-term trends.”
There are other replies in Nature critical of the Boyce, et al. thesis that phytoplankton is declining, including (Mackas 2011) and (Rykaczewski and Dunne 2011). The extensive criticism led Boyce, et al. to publish two follow up papers, one in 2012 in Limnology and Oceanography Methods (Boyce, Lewis and Worm 2012) and another in 2014 in Progress in Oceanography (Boyce, et al. 2014). Thus, we appear to have conflicts between different datasets. Another discussion, regarding data in the Indian Ocean is very similar. A paper published in Geophysical Research Letters (Roxy, et al. 2016) states:
“Earlier studies had described the western Indian Ocean as a region with the largest increase in phytoplankton during the recent decades. On the contrary, the current study points out an alarming decrease of up to 20% in phytoplankton in this region over the past six decades. We find that these trends in chlorophyll are driven by enhanced ocean stratification due to rapid warming in the Indian Ocean, which suppresses nutrient mixing from subsurface layers. Future climate projections suggest that the Indian Ocean will continue to warm, driving this productive region into an ecological desert.”
OK, earlier studies show the largest increase in phytoplankton ever recorded, but our study finds an “alarming” decrease due to warming that will turn the Indian Ocean into a desert if the global climate models are correct. We sense some ambiguity and uncertainty, perhaps a closer look at the data and measurement methods is warranted. After all, it is just barely possible that the data isn’t good enough or definitive enough to draw such extraordinary conclusions. We are confident that the climate model projections are not very accurate.
Ocean phytoplankton overview
The ancestors of photosynthetic eukaryotes or the larger variety of phytoplankton evolved over 1.5 billion years ago and radically changed the world by consuming carbon dioxide and increasing the oxygen content of the atmosphere from nearly zero to 20% by the beginning of the Cambrian 540 million years ago. The first photosynthesizing organisms were probably very similar to modern cyanobacteria, a primitive variety of ocean dwelling phytoplankton. Cyanobacteria have no shell, are very small, and may be the origin of photosynthetic organelles in plant cells called chloroplasts. All phytoplankton live in the upper layers of the ocean in the “euphotic zone” or the part of the ocean that has enough light from the Sun for photosynthesis. Photomicrographs of some larger phytoplankton are shown in figure 1 from a paper in Science in July 2004 (Falkowski, et al. 2004) called “The Evolution of Modern Eukaryotic Phytoplankton.”
Figure 1, Phytoplankton photomicrographs. A: Diatom chain, B: a diatom valve, C: a tropical coccolithophore, D: a pair of phycomas, E: a clump of coccospheres, F: an athecate dinoflagellate, G: a thecate dinoflagellate. Source: (Falkowski, et al. 2004).
The early phytoplankton were very different from the species that dominate the oceans today. Modern phytoplankton evolved during the Mesozoic some 250 million years ago and are comprised of three large related types (or “clades”), the dinoflagellates, coccolithophores and diatoms shown in Figure 1. These clades all contain plastids of the original red algae and are descended from it.
Considerable doubt exists that humans are changing the climate, but there is no doubt that phytoplankton radically changed the world, as well as the climate, and most would say for the better. So why do we have so many very different interpretations of the data on phytoplankton abundance in the oceans? Phytoplankton are extremely important to the living Earth, the primary source of oxygen, the primary sink for CO2, and the ocean’s primary food source. If they are on a permanent declining trend, it is a big deal. We will frame the discussion by listing the objections to the original 2010 paper by Boyce, et al. and will follow these with the related comments from Boyce, et al. 2014.
Blending Bias, Mackas, 2011, Nature
David Mackas of the Institute of Ocean Sciences, Fisheries and Oceans in British Columbia, Canada (Mackas 2011) has reported in Nature that much, if not most, of the decline reported by Boyce, et al. (2010) is due to a bias created by blending the two data types they used. Boyce et al. pooled estimates of chlorophyll concentration from Secchi disks (Figure 2) and direct measurements from in situ profiles.
Secchi disks were invented in 1865 by Father Pietro Angelo Secchi to measure the clarity of the Mediterranean Sea. They are a 20 cm disk painted in bright contrasting colors. The ship’s crew lowers the disk into the ocean and notes the depth at which the disk becomes invisible. It was a standard measurement made by ships for nearly 100 years. In Figure 2 a Secchi disk is lowered from a ship on a 1949 voyage.
Figure 2. A photograph of a ship’s crew lowering a Secchi disk into the ocean in 1949 to measure water transparency. The image is from The Art Archive/R. Sisson/NGS Image Collection and copied from (Siegel and Franz 2010).
The in-situ measurements used by Boyce et al. are shipboard measurements of total chlorophyll pigment concentration. These measurements have been done on surface samples using spectrophotometry, fluorometric analysis or in vivo measurements of phytoplanktonic fluorescence over the past 100 years. Boyce and colleagues found a correlation between the in situ and Secchi disk measurements of the total chlorophyll pigment concentration. To do so they had to take the logarithm (base 10) of the concentrations first, the correlation coefficient of the logarithm of the concentrations was 0.77 for 13,700 samples. Since both measurements are in mg/m3 it is unclear why taking the logarithm was necessary, except to improve the correlation coefficient. The comparison is shown in Figure 3.
Figure 3. A plot comparing the Log10 Secchi disk derived total chlorophyll concentration to the Log10 in-situ Chlorophyll concentration and a map of the difference of the logarithms. Note a difference of one is an order of magnitude or a difference of 10 times. Source Boyce, et al. supplementary materials, figure S2.
In the text of the article Boyce, et al. report the correlation coefficient between the two concentration measures as 0.52, it rises to 0.77 after the log10 is taken of both concentration estimates. The logarithmic scaling reduces the apparent bias in that the Secchi disk estimates are biased high to the in-situ measurements by 25-50% throughout most of the measurements. Since the earlier measurements are mostly Secchi disk measurements and later measurements are mostly in-situ, the amount of bias is a function of time and this bias accounts for much of the one percent change per year documented by Boyce, et al. according to David Mackas (Mackas 2011).
In response to the criticism by Mackas and others (see below) Boyce, Lewis and Worm (2012) developed a new more comprehensive global database of chlorophyll concentration from 6 meters to 20 meters depth. The database uses a lot of data, including Secchi disk data, and is compared to satellite data but does not use satellite data. Besides limiting the depth range to 6 meters to 20 meters, they also do not include any measurements made in water with a total depth of less than 20 meters or within 1 km of a coast line. Phytoplankton exist in the top 100 to 300 meters of the ocean (Woods Hole Oceanographic 2018) so the volume investigated by Boyce, et al. is very small relative to the total volume of ocean containing phytoplankton.
In the 2012 database paper, Boyce and colleagues address this criticism by first calibrating all sources of chlorophyll concentration to a quality-controlled set of in situ measurements. This is an attempt to makes the various trends of measurements match, eliminating the systematic bias of the 2010 study.
Rykaczewski and Dunne (2011) Nature
This critique of Boyce, et al. (2010) is similar to Mackas (2011) but more detailed. Ryaczewski and Dunne examine the bias described by Mackas geographically and noticed that the bias was only 5% at low concentrations and >100% at high concentrations. Some of this trend may be due to taking the logarithm of the concentrations to build the function and then the anti-log to get the result. They also found that the bias was much higher in the Northern Hemisphere than in the Southern Hemisphere and that the pattern of the bias matched the pattern of the decline reported by Boyce, et al., see figure 4.
Figure 4. A comparison of the computed bias to the Boyce, et al. decline.
As we can see in Figure 4, the expected pattern of the bias is very similar to the pattern of decline in phytoplankton reported by Boyce, et al. in 2010. Rykaczewski and Dunne conclude, quite reasonably, that the reported decline is due to bias.
The critique of (McQuatters-Gollop, et al. 2011)
Probably the most devastating critique of Boyce, et al. (2010) was by McQuaters-Gollop, et al. (2011). Regarding the claim by Boyce, et al. that there has been a 1% decline per year in phytoplankton biomass, they wrote that Continuous Plankton Recorder (CPR) surveys in the Atlantic and Pacific show an increase in an index of chlorophyll, the Phytoplankton Color Index (PCI), as discussed in a quote from their paper in the introduction.
They found that more than 5 million nautical miles of ocean have been sampled by ships towing the CPR and more than 250,000 phytoplankton and zooplankton samples have been analyzed using a virtually unchanged methodology over the last 80 years and this data suggests that the concentration of phytoplankton in the oceans has increased. Figure 5 shows the Atlantic Phytoplankton Color Index data (PCI) from numerous CPR surveys.
Figure 5. The CPR surveys of phytoplankton concentration in PCI units for the North Atlantic.
Figure 5 shows that phytoplankton is increasing in the North Atlantic, precisely the opposite of the trend suggested by Boyce, et al. (2010). McQuaters-Gollop, et al. (2011) also found that if the Secchi disk measurements are removed from the Boyce, et al. database, the negative trends in the Atlantic and Pacific oceans reverse and become positive trends.
Boyce, et al. 2014
Boyce and colleagues reviewed all the criticism of their 2010 paper and, as a result, created a new database that is presented in their 2012 paper (Boyce, Lewis and Worm 2012). In the 2014 paper in Progress in Oceanography (Boyce, et al. 2014) they provide a new estimate of the trend in ocean phytoplankton concentration. Their new estimate, including the questionable Secchi disk measurements, showed a worldwide decline of 0.23% per year, about a quarter of the decline shown in the 2010 paper. In their global grid, 57% of the cells showed a decline in phytoplankton in the depth range studied, 6 meters to 20 meters.
It is significant that the weighted mean decline rate of the in situ measured data is significantly different from the Secchi disk decline rate for all trends and for statistically significant trends. This suggests, as McQuaters-Gollop (2011) noted that including the questionable Secchi disk data is critical to producing a declining trend. The differing distributions are apparent in Figure 6, from Boyce, et al. (2014).
Figure 6. Probability distributions of the logarithm of the Secchi disk concentrations in blue and the logarithm of the in situ measurements in red from all available data. This suggests either an actual change, since the early measurements are mostly Secchi disk estimates, or that the change seen in the study is dependent upon the distribution of Secchi disk estimates, source (Boyce, et al. 2014).
Let us establish some basics about phytoplankton. It is important to remember that phytoplankton exists from the surface of the ocean to 200 meters or deeper over most of the world ocean, which covers 70% of the Earth’s surface. Some chlorophyll abundance and productivity depth profiles are shown in Figure 7.
Figure 7. Depth profiles for total phytoplanktonic carbon (Phyto C) primary production, phytoplanktonic production (PROD) and microzooplankton grazing (GRAZ) in the Costa Rica Dome area in July 2010, these measurements are 24-hour measurements from (Landry, et al. 2016).
In the Costa Rican Dome, phytoplankton are productive down to nearly 100 meters water depth, in other parts of the world they are productive to nearly 200 meters, sometimes even deeper. In Figure 8 we see total chlorophyll a, phytoplankton carbon biomass, and phytoplankton growth rate as a function of depth in offshore Hawaii. Significant growth is taking place at a depth of 140 meters.
Figure 8. Mean depth profiles for total chlorophyll a, carbon biomass, and phytoplankton growth rate offshore Hawaii from (Landry, Brown, et al. 2008). The “IN” profiles were conducted in an eddy being studied and the “OUT” profiles are the control outside of the eddy.
In Figure 8 it is significant that total chlorophyll and phytoplanktonic biomass can peak below 100 meters. As noted above, Boyce, et al., in both papers, only collected and used data from 6 meters depth to 20 meters depth. Even if they successfully sampled a significant portion of these depths over the entire world, it would only account for 9% of the euphotic zone, assuming an average depth of 150 meters. In many areas, such as Hawaii, they wouldn’t even sample the primary production zone.
Roxy, et al. (2016) report a decline in phytoplankton in the Indian Ocean, but readily admit that previous studies showed an increase. Every study shows that climate changes increase phytoplankton at some depths and decrease it in others. The phytoplankton will find an optimum depth where they maximize both the nutrient concentration and the available light from Sun. The best nutrient sources are in deeper water, and the best sunlight is shallower. The optimum depth changes with climate, season and weather. Sampling a thin layer from 6 meters to 20 meters, with the subject of your study moving up and down, often well outside that layer, due to climate change, season and weather, in an unpredictable fashion, is not ideal.
I appreciate the hard work of all these researchers, but frankly after reading numerous papers, I don’t think they have any idea if total plankton is increasing or decreasing or staying the same. They are sampling a huge ocean with a metaphorical tea cup. According to Boyce, et al. a record of plankton abundance of as many as 40 years is required to separate long-term natural trends from short term fluctuations (Boyce, et al. 2014). Given that the Atlantic Multidecadal Oscillation is about 60 years, more time than that may be required. Further, current high quality in situ data is very sparse and Secchi disk data is problematic. Satellite data is available for less than 30 years and very much affected by weather conditions and a shallow depth of investigation. Satellite data cannot distinguish between vertical movements of the phytoplankton population and their overall abundance. To make matters worse, different researchers using the same satellite data find plankton both increasing and decreasing globally (Boyce, Lewis and Worm 2010). Satellite data may never be adequate for this purpose. In short, this is an interesting and important scientific question, but the data we have available today is clearly unable to answer it.
Before I investigated this issue, I had seen Boyce, et al. 2010 cited in media reports many times. It is significant that I have not seen the rebuttals to their paper cited in media reports. Thus, a catastrophic decline in phytoplankton is reported, but extremely important, and valid, criticisms of the study are not worth mentioning? The media are a very poor source of information on science.
Boyce, D. G., M. R. Lewis, and B. Worm. 2010. “Global phytoplankton decline over the past century.” Nature 466. https://www.nature.com/articles/nature09268.
Boyce, D., M. Lewis, and B. Worm. 2012. “Integrating global chlorophyll data from 1890 to 2010.” Limnology and Oceanography Methods 10: 840-852. http://onlinelibrary.wiley.com/doi/10.4319/lom.2012.10.840/full.
Boyce, Daniel, Michael Dowd, Marlon Lewis, and Boris Worm. 2014. “Estimating global chlorophyll changes over the past century.” Progress in Oceanography 122: 163-173. https://www.sciencedirect.com/science/article/pii/S0079661114000135.
Falkowski, Paul, Miriam Katz, Andrew Knoll, Antonietta Quigg, John Raven, Oscar Schofield, and F. Talor. 2004. “The Evolution of Modern Eukaryotic Phytoplankton.” Science 305 (5682): 354-360. http://science.sciencemag.org/content/305/5682/354.full.
Fehling, Johanna, Keith Davidson, Christopher Bolch, Tim Brand, and Bhavani E. Narayanaswamy. 2012. “The Relationship between Phytoplankton Distribution and Water Column Characteristics in North West European Shelf Sea Waters.” PLOS. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3313960/.
Landry, Michael, Karen Selph, Moira Decima, Andres Gutierrez-Rodriguez, Michael Stukel, Andrew Taylor, and Alexis Pasulka. 2016. “Phytoplankton production and grazing balances in the Costa Rica Dome.” Journal of Plankton Research 38 (2): 366-379. https://academic.oup.com/plankt/article/38/2/366/2375234.
Landry, Michael, Susan Brown, Yoshimi Rii, Karen Selph, and Robert Bidigare. 2008. “Depth-stratiﬁed phytoplankton dynamics in Cyclone Opal, a subtropical mesoscale eddy.” Deep Sea Research II 55: 1348-1359. https://s3.amazonaws.com/academia.edu.documents/46414156/Depth-stratified_phytoplankton_dynamics_20160612-6046-1f7cvb2.pdf?AWSAccessKeyId=AKIAIWOWYYGZ2Y53UL3A&Expires=1521056786&Signature=DqsVgm7TdQA0WliY0yP0x6VwLy0%3D&response-content-disposition=inline%.
Mackas, David. 2011. “Does blending of chlorophyll data bias temporal trend?” Nature 472: e4-e5. http://imedea.uib-csic.es/master/cambioglobal/Modulo_V_cod101612/articulos%20para%20presentaciones_escoger/Boyce_etal_2011%20comments.pdf.
McQuatters-Gollop, Abigail, Philip Reid, Martin Edwards, Peter Burkhill, Claudia Castellani, Sonia Batten, Winfried Gieskes, et al. 2011. “Is there a decline in marine phytoplankton?” Nature 472. https://www.nature.com/articles/nature09950.
Roxy, Mathew Koll, Aditi Modi, Raghu Murtugudde, Vinu Valsala, Swapna Panickal, S. Kumar, M. Ravichandran, M. Vichi, and M. Levy. 2016. “A reduction in marine primary productivity driven by rapid warming over the tropical Indian Ocean.” Geophysical Research Letters 43 (2): 826-833. http://onlinelibrary.wiley.com/doi/10.1002/2015GL066979/full.
Rykaczewski, Ryan, and John Dunne. 2011. “A measured look at ocean chlorophyll trends.” Nature 472. http://imedea.uib-csic.es/master/cambioglobal/Modulo_V_cod101612/articulos%20para%20presentaciones_escoger/Boyce_etal_2011%20comments.pdf.
Siegel, David, and Bryan Franz. 2010. “Oceanography: Century of phytoplankton change.” Nature 466: 569-571. https://www.nature.com/articles/466569a.
Woods Hole Oceanographic. 2018. “Phytoplankton.” Know your Ocean. http://www.whoi.edu/main/topic/phytoplankton.