Since the start of the industrial revolution our carbon dioxide output has increased dramatically. The oceans absorb a large portion of the atmospheric CO2 and therefore reduce the greenhouse effect. The benefits are obvious but today we know that there are major downsides. The CO2 absorbed by the oceans leads to changes in the seawater chemistry, some with major implications for marine organisms.
A visual representation of the oceans carbonate system with its buffer reactions
Calcium carbonate shells from pteropods were placed in seawater with a pH of 7.8 (the projected pH for 2100). The top row shows the shells before the experiment, the bottom row shows the shells after 45 days of exposure.
In the last century, a decline in ocean pH from 8.2 to 8.1 has been documented. Since the pH scale is logarithmic , this decline represents a 30 % increase in acidity. This has dramatic impacts on shell building organisms. A lower pH translates into increased rates of dissolution of calcium carbonate and a lower amount of free carbonate ions in the water. While the CO2 cycle in the oceans is understood fairly well, there are many knowledge gaps in the implications for marine organisms.
Book: Introduction to Oceanography (Webb). (2019, September 16). Roger Williams University. https://geo.libretexts.org/@go/page/4449
In the following sections, you will find examples of my work. It does NOT contain all the code necessary for the project.
Let's start out easy with some typical data wrangling processes. First we need to load some data, filter it for our study area and remove those pesky NAs! We also seperate our study area in two zones to counteract the effect of different sampling efforts.
Remember TS plots?
We can separate water masses!
You need carbon system variables such as pH, total alkalinity, or dissolved inorganic carbon BUT you lost most of them?! Fear not! You just need two of them to compute the rest. Isn't that amazing?
Data is nothing without errors... so let's compute them!
After some more data wrangling, we can finally compute a robust linear model. In this case we compute the model for pH data. What do we need? The mean pH per year and the mean error per year.
After all this code, finally some results.
Presented here is the pH trend between 1971 and 2016 in subsurface waters (100 m - 500 m) of the Portuguese Margin. The measurements are coloured according to their source, GLODAP in red, ICES in green, and old cruise data in blue. The black points represent the means of each year. The pH value decreased at a steady rate of -0.0017 ± 0.0002 per year. The associated p-value is < 0.05. Through adding the pH values from the "rescued" data, we were able to fill knowledge gaps and increase the time frame of the linear model by about one decade.
You still have not enough of R code? Are you curious about the intermediate steps of the script and want to see more trends for xcCO3 and anthropogenic carbon? You are lucky, feel free to explore more.