From the Southern Ocean to My Spreadsheet

2006 example of the side-by-side Salp vs. Tmax maps on the PAL LTER sampling grid

The Palmer Antarctica Long Term Ecological Research (PAL LTER) cruises are crazy busy! The study area is organized into lines of sampling stations every 100 nautical miles down the Western Antarctic Peninsula. No matter the day or time when the ARSV Laurence M. Gould reaches the next station, it is time to collect data. I was working shifts from midnight to noon. At each sampling station, the oceanographer’s workhorse – the CTD – is sent down nearly to the seafloor. This device records salinity (calculated from Conductivity) and Temperature as a function of Depth for the entire water column. On the way back up, bottles are snapped shut so that researchers can analyze water samples from various depths. During the 21 PAL LTER cruises from 1993 to 2013 there were more than 2,000 CTD casts. As soon as the CTD is back on deck, it was time for Team Zooplankton to get to work. We would first deploy a net with a 2-m2 opening off the back deck and send it down to a depth of 300 meters. After getting the sample onboard, we identified and counted all of the cool critters that had floated into the net. Once we finished processing the tows it was usually time to throw the net back into the ocean. These 2-m2 net tows gave me more than 1,100 data points on Salpa thompsoni abundances for the time series.

Recovering the 2-m plankton net

With data aplenty I began picking out the information that would help explain if/how warm UCDW was influencing the WAP salp population. The maximum sub-surface temperature (Tmax) was my signal for UCDW, so I needed to compare those numbers directly to Salpa thompsoni densities. After some MATLAB assistance from data wizard Joe Cope, I eliminated all CTD casts without a usable Tmax  (usually at shallow, nearshore stations). I then paired individual Tmax values to salp densities by selecting the CTD casts and net tows nearest to one another time and space. All paired tows and CTD casts were carried out within 10 nautical miles and 24 hours of each other, normally much closer than that. This left me with 880 comparable data points over 21 different January cruises. I used the skills I learned in one of my favorite college classes – Introduction to Geographic Information Systems – to plot these temperature and salp values side by side on maps of the study region.

Recovering the CTD rosette

Although I study biology, the salp data was more challenging for me to interpret than the temperature numbers from the get-go. Nearly every Tmax measurement fell between zero and two degrees Celsius. I could easily set cut-offs at 1.6 and 1.7 degrees to select stations where UCDW was present. The salp density data was another beast entire. Almost half of the tows did not catch any salps and the values ranged from 0.1 to over 2,000 individuals per 1,000m3. Fortunately PhD student Josh Stone came to my rescue, because he faced a similar situation when studying salps off of Bermuda. To deal with this heavily skewed data, I replaced all zeroes with the lowest non-zero value and subsequently took the natural log of the density values. This made it much easier to perform statistical tests and to represent the densities graphically. Now that I have massaged the data, the trends still aren’t jumping out of the screen at me. It’s time to run statistical tests and look closer for patterns through space and time.