Oceanography | Vol. 38, No. 2
78
fluorometers, we will focus here only on the discrete under
way Chl-a data. Before collection, the date and time of sam
pling were recorded along with the current fluorometer read
ings. Fifteen to 20 samples were collected on a random timeline
during the cruise, while ensuring collection of half the samples
during the day and about half during the night, to capture the
effects of nonphotochemical quenching (Marra, 1998; Holm-
Hansen et al., 2000). Additional effort was devoted to maximiz
ing the dynamic range of fluorescence and corresponding Chl-a
concentrations, based on real-time observations of the under
way fluorescence signals (e.g., during periods of unusually low
or high fluorescence). A collection container with a volume
between 500 mL and 1 L was rinsed three times with underway
water, then filled. Three plastic 152 mL bottles were then filled to
the top with the underway water in triplicates. Immediately after
collection, the entire volume of each triplicate was filtered onto
Whatman GF/F 25 mm filters using gentle vacuum (not exceed
ing 150 mm Hg) in a light-limited environment to avoid any
degradation of the Chl-a pigments. The filters were then each
placed in glass tubes containing 6 mL of 95% ethanol, capped,
and stored in the dark at room temperature to extract the Chl-a
for approximately 12 hours (±2 hours). After the extraction
period, the fluorescence of the samples was recorded with a
Turner 10AU fluorometer first as is, and then with the addition
of acid to correct for phaeopigments (Wasmund et al., 2006).
The discrete underway sample data were digitized and orga
nized, then Chl-a concentration, in mg m–3 (= μg L–1), was cal
culated using coefficients obtained from the in-lab fluorometer’s
calibration; this was performed before each cruise based on pure
Chl-a standards (Sigma-Aldrich, from Anacystis nidulans algae).
Each data point was then given a quality flag based on the IODE
(International Oceanographic Data and Information Exchange)
quality flag scheme (IOC, 2013) so that only the highest quality
data would be included in the post-calibration. Discrete under
way Chl-a data from six NES-LTER cruises are available on the
EDI data portal (Menden-Deuer et al., 2022).
SECTION 2. USING DISCRETE CHL-a TO
POST-CALIBRATE SENSOR-BASED FLUORESCENCE
(1.5 hours)
There can be substantial differences between manufacturer-
calibrated continuous fluorescence data and discrete Chl-a con
centrations. Manufacturer-calibrated fluorescence values con
verted to Chl-a concentrations (mg m–³) should be interpreted
with caution because the calibration is typically performed
using either pure Chl-a extracts or single-species phytoplank
ton cultures that may not accurately reflect the local phyto
plankton community, environmental conditions (e.g., tempera
ture), or optical properties encountered in the field. Factors
such as species composition, physiological state, light his
tory, and colored dissolved organic matter (CDOM) can all
influence the fluorescence signal independent of actual Chl-a
concentration. The optical components of the fluorescence sen
sor may also be biofouled during deployment. Although this
is minimized by cleaning the sensors before and after each
deployment and by maintaining a high flow rate, any biofoul
ing can still alter the recorded optical signal. As a result, with
out cross-validation, these manufacturer-derived values can be
substantially different from in situ Chl-a. Therefore, it is crucial
to acknowledge, correct for, and interpret the uncertainty and
imprecision in in vivo fluorescence data to interpret the fluo
rescence signal (Cullen, 1982; Falkowski and Kiefer, 1985; Xing
et al., 2017). To obtain reliable, accurate, high-resolution Chl-a
data from in vivo fluorescence, the continuous fluorometer data
must undergo post-calibration against discrete Chl-a values.
The steps required for this data management are the subject of
this hands-on exercise.
PART 1. PLOTTING SENSOR-BASED CHL-a FLUORESCENCE
VS. EXTRACTED CHL-A CONCENTRATIONS
Goals. Linking sensor-based underway chlorophyll-a (Chl-a)
fluorescence and discrete Chl-a data. Introduce methods required
for post-calibrating sensor-based Chl-a fluorescence data.
Expected Outcomes. Develop familiarity with linear regres
sion, including the concepts of slope, intercept, and coefficient
of determination. Understand the significance of linear regres
sion results and their application in post-calibrating underway
Chl-a fluorescence data.
Narrative. It is now time to compare the discrete Chl-a concen
trations with the corresponding underway fluorescence values
observed when sampling (Figures 3 and 4). The goal here is to
identify whether both fluorometers are equally well suited to use
for the post-calibration and to identify the coefficients that will
FIGURE 3. Raw underway fluorescence (mg Chl-a m–3) during the EN661
NES-LTER transect cruise (February 3 to February 7, 2021, winter in the
Northern Hemisphere) from the WetStar (dark green) and the ECO-Fl (light
green) fluorometers. The discrete Chl-a concentrations collected in tripli
cate during the cruise are represented by black dots. Only discrete Chl-a
data with an IODE Quality Flag = 1 (good) are shown. Note the change in
fluorescence from the WetStar fluorometers on 02/05/2021, which cor
responds to the change observed after the cleaning of the instruments.