Guest Post by Willis Eschenbach
Following on from Anthony’s article, here are my thoughts about the phytoplankton paper “Global phytoplankton decline over the past century”, by Daniel G. Boyce, Marlon R. Lewis & Boris Worm.
I started to write about this earlier, but I decided to wait until I had the actual paper. The paper in question is behind a paywall at Nature Magazine, but through my sub-oceanic channels (h/t to WS) I have obtained a copy. The paper makes two main claims, that: a) the numbers of phytoplankton have been cut by more than half since 1900, and b) the general warming of the global oceans is the reason for the declining numbers of phytoplankton.
First, what are phytoplankton when they are at home, and where is their home? Plankton are the ubiquitous soup of microscopic life in the ocean. Phytoplankon are the plant-like members of the plankton, the ones that contain chlorophyll and feed on sunshine. Phytoplankton are to the ocean what plant life is to the land. Almost all oceanic life depends on phytoplankton. Other than a thin strip of seaweeds and sea grasses along the coasts, phytoplankton are the microscopic plants that are the foundation of the vast entire oceanic food chain. Without phytoplankton there would be no deep water oceanic life to speak of. Figure 1 shows where you find phytoplankton:
Figure 1. Global distribution of phytoplankton. Lowest concentration is purple and blue, middle concentration is green, highest concentration is yellow and red. Source http://www.nasa.gov/vision/earth/environment/0702_planktoncloud.html
So where did the Nature paper go wrong?
The short answer is that I don’t know … but I don’t believe their results. The paper is very detailed, in particular the Supplementary Online Information (SOI). It all seems well thought out and investigated … but I don’t believe their results. They have noted and discussed various sources of error. They have compared the use of Secchi disks as a proxy, and covered most of the ground clearly … and I still don’t believe their results. Here’s exactly why I don’t believe them.
This is their abstract (emphasis mine):
In the oceans, ubiquitous microscopic phototrophs (phytoplankton) account for approximately half the production of organic matter on Earth. Analyses of satellite-derived phytoplankton concentration (available since 1979) have suggested decadal-scale fluctuations linked to climate forcing, but the length of this record is insufficient to resolve longer-term trends.
Here we combine available ocean transparency measurements and in situ chlorophyll observations to estimate the time dependence of phytoplankton biomass at local, regional and global scales since 1899. We observe declines in eight out of ten ocean regions, and estimate a global rate of decline of ~1% of the global median per year. Our analyses further reveal interannual to decadal phytoplankton fluctuations superimposed on long-term trends. These fluctuations are strongly correlated with basin-scale climate indices, whereas long-term declining trends are related to increasing sea surface temperatures. 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.
The first clue to where they went wrong is visible in Fig. 1. Although as you can see there is more phytoplankton in the cooler regions of the north, the same is not true in the corresponding regions in the south despite the ocean temperatures being very similar. In addition, there are many places where the ocean is warm (e.g. tropical coasts) that have lots of phytoplankton, while in other warm areas there is very little phytoplankton.
The rude truth of phytoplankton is this: phytoplankton growth is generally not limited by temperature. Instead, it is limited by nutrients. Where nutrients are plentiful, the phytoplankton grow regardless of temperature. Nutrients are more common along the coastline, where sub-oceanic currents come to the surface bringing nutrients from the deep ocean floor, and rivers bring nutrients from inland. For example, in Fig. 1 you can see the nutrients from the Amazon river causing the red area at the river mouth (north-east South American coast).
Indeed, the fact that phytoplankton are generally nutrient limited rather than temperature limited has been demonstrated in the “ocean fertilization” experiments using rust. If you spread a shipload of rust (iron oxide) out into the tropical ocean, you generally get an immediate bloom of phytoplankton. Temperature is not the problem.
So to start with, the idea that increasing temperature automatically leads to decreasing phytoplankton is not generally true. There are vast areas of the ocean where higher temperatures are correlated with more phytoplankton. For example, the warmer deep tropics generally have more phytoplankton than the cooler adjacent subtropics.
The paper’s most unbelievable claim, however, is their calculation that since 1899, the density of phytoplankon has been decreasing annually by 0.006 milligrams per cubic metre (mg m-3). They give the current global density of phytoplankton as being 0.56 mg m-3. Thus they are claiming that globally the concentration of phytoplankton has dropped by more than 50% over the last century.
Now, a half century ago I learned to sail on San Francisco Bay. Since then I’ve spent a good chunk of my lifetime at sea, as a commercial fisherman from California to the Bering Sea, as a sailboat delivery crewman, as a commercial and sport diver, and as a surfer. And call me crazy, but I simply don’t believe that the sea only has half the phytoplankton that it had in 1900. If that were true, it would not take satellites and complex mathematical analysis to show it. People would have noticed it many years ago.
I say this because phytoplankton are the base of almost the entire mass of oceanic life. They are what almost all other life in the ocean ultimately feeds on, predators and prey as well. The authors of the study do not seem to realize that if the total amount of phytoplankton were cut by more than half as they claim, the total mass of almost all living creatures in the open ocean would be cut about in half as well. No way around it, every farmer knows the equation. Half the feed means half the weight of the animals.
And I see no evidence of that having happened over the last century. It certainly does not accord with my own extensive practical experience of the ocean. And I see no one else making the claim that we only have half the total mass of deep-water oceanic life that we had a century ago..
The other thing that makes their claimed temperature/phytoplankton link very doubtful is that according to the HadISST dataset, the global ocean surface temperature has only increased by four tenths of a degree C in the last hundred years.
Four tenths of a degree … an almost un-noticeable amount. Yet their paper says (emphasis mine):
Our analyses further reveal interannual to decadal phytoplankton fluctuations superimposed on long-term trends. These fluctuations are strongly correlated with basin-scale climate indices, whereas long-term declining trends are related to increasing sea surface temperatures.
These kinds of claims drive me nuts. Is there anyone out there that truly believes that a change of global average ocean temperature of four tenths of a degree C over the last hundred years has cut the total mass of phytoplankton, and thus the total mass of all oceanic creatures, in half? Really?
So that’s why I say I don’t know where their math went wrong, but I don’t believe their results. I don’t believe we’ve lost about half the total mass of all oceanic creatures. Half the planet’s open ocean dwellers? Where is the evidence to support that outrageous claim? And I don’t believe that an ocean temperature change of four tenths of a degree over a century has made much difference to phytoplankton levels, as they grow at all temperatures.
Why don’t I know where their math went wrong? Unfortunately, they have not posted up the data that they actually used. Nor have they shown any of their data in the form of graphs or tables. Instead, they have shown model results, and merely pointed to general websites where a variety of datasets are maintained. So we don’t know, for example, whether they used the 1° grid version or the 2.5° grid version of a given dataset. Nor have they posted the computer code that they used in the analysis. Plus, the very first link in their paper to the first and most important data source is dead.
Grrrr … but dead link or not, pointing to a website as the data source in their kind of paper is meaningless. To do the analysis, they must have created a database of all of the observations, with the meta data, and the details for the type etc. for each observation. If they would include that database and their code in the SOI, then someone might be able to figure out where their math went wrong … my guess is that it may be due to overfitting or misfitting of their GAM model, but that’s just a wild guess.
It is a shame that they did not post their data and code, because other than the lack of data and code it is a fascinating analysis of a very interesting dataset. I don’t accept their analysis of the data because it doesn’t pass the “reasonableness” test, but that doesn’t mean that the dataset does not contain valuable information.
[Update] An alert reader noted that the image in Figure 1 was of a particular month and not a yearly average. So I’ve made a short movie of the variations in plankton over the year.