Questioning High Nitrogen Fixation Rate Measurements in the Southern Ocean

Full text

matters arising
https://doi.org/10.1038/s41561-021-00873-3
1Department of Oceanography, University of Hawai’i at Manoa, Honolulu, HI, USA. 2Department of Marine Sciences, University of Connecticut, Groton,
CT, USA. 3Ocean Sciences Department, University of California Santa Cruz, Santa Cruz, CA, USA. ✉e-mail: aewhite@hawaii.edu
Nitrogen fixation in the global ocean is a microbially mediated pro-
cess performed by a special class of organisms able to enzymatically
reduce dinitrogen gas (N2) to ammonia to fuel cellular nitrogen (N)
demands1–3. There now seem to be few ocean habitats wholly unfa-
vourable to at least a subset of N2-fixing organisms, and, in this con-
text, Shiozaki et al.4 recently reported substantial nitrogen fixation
in one of the coldest and most isolated regions of our planet: the
coastal waters near the ice edge of Antarctica5,6. The authors specu-
lated that iron additions from melting sea ice fuel N2 fixation activ-
ity of the unicellular cyanobacteria UCYN-A/haptophyte symbiosis
and argue that the results suggest that “marine nitrogen fixation is
a ubiquitous process in the global ocean, and that UCYN-A is the
keystone species for making it possible”.
We read this Article with great interest given the very high rates
of oceanic N2 fixation in such a cold (below 0 °C), nitrate-rich
(>15 μmol l−1) habitat. We were immediately struck by the magni-
tude of the rate of 44 nmol N l−1 d−1 reported at the ice edge4. This
rate is extraordinarily high; compared with a global compilation by
Luo et al.7, it is in the highest 1% of oceanic rates ever measured (see
fig. 1 of Luo et al.7 and data in ref. 8). The authors thus stated that the
“study sheds light on nitrogen fixation as an alternative source of
[reactive N] to support primary production in the Antarctic Ocean”
against a backdrop of micromole per litre concentrations of nitrate.
This is a claim that compels scrutiny of the data.
Upon examination of data generously shared with us by Shiozaki
et al. (not included in the published Article), we were struck by
three salient points.
Replicate measurements from Station E were treated
differently
The reported rate at Station E of 44 nmol N l−1 d−1 in sea ice derives
from an outlier. All rate determinations yielded low or undetect-
able rates of N2 fixation (<2 nmol N l−1 d−1) with this one exception.
In the methods, it is stated that replicate measurements of perti-
nent rate-specific terms were conducted4. These terms include the
inherent nitrogen isotopic composition of particulate material (the
atomic per cent (at.%) of particulate nitrogen, PN, before tracer
addition, APN–T0), the N isotopic composition of added tracer (at.%
of added 15N2, AN2), the incubation period (Δt), and the final N iso-
topic composition after incubation (at.% of PN at termination of
experiment, APN–Tf), all of which are necessary to accurately quantify
the nitrogen fixation rate (NFR)9, as described in Shiozaki et al.4 and
derived by Montoya et al.9:
NFR =
(A
PN−Tf −
A
PN−T0
)
(
AN2
−
APN
−
T0
)×
[PN]
Δt(1)
The data shared with us confirm that all of the above terms were
measured (which is commendable, as this is not always done10)
and rate-specific terms were indeed calculated from replicates
at all stations except Station E (the ice-edge station, see Table 1).
The reported high rate of 44 nmol N l−1 d−1 was calculated using
one of three replicate measurements of APN–Tf (0.450%), despite the
remaining replicates having a relatively low APN–Tf (0.368 ± 0.005%).
That lone high APN–Tf value is notably ~6 s.d. from the mean of
APN–Tf values achieved in incubations at other stations where rates
were detectable (n = 16, 0.368 ± 0.002%; Table 1). As noted above,
rate measurements at all other stations were uniformly low (0.2–
1.9 nmol N l−1 d−1), and all were estimated from extant replicates at
the exclusion of none.
It is not clear why reasonable replicates that were consistent with
all other data were discarded, with the exception of one outlier.
There are several reasons for a singularly high APN–Tf, including sam-
ple contamination during collection or during storage and process-
ing of samples, carryover in the mass spectrometer or nonlinearities
of the mass spectrometer. All of these potential explanations seem
more plausible than excluding replicates, and we remain concerned
that key underlying data were not presented, and that the implica-
tions of the lack of reproducibility at Station E were not discussed.
We also note that the at.% increase during the incubations was rela-
tively small, with a median value of 0.0012% (that is, 3‰ versus
air), which could easily arise from inherent N cycling (for example,
see ref. 11), rather than the incorporation of the 15N2 tracer. Control
incubations would have helped address this issue but were not per-
formed. Moreover, the median increase was of the same order as
three times the standard deviation of initial APN–T0 values (0.002%),
thus at (or below) a standard definition of the limit of detection for
this method10.
Conspicuously low UCYN-A gene abundance
The reported UCYN-A gene abundance at Station E is far too low to
account for the corresponding rate of 44 nmol N l−1 d−1. Gene abun-
dances of the nitrogenase gene marker for UCYN-A (the proposed
driver of these rates) are estimated to be 129 gene copies per litre,
and never higher than the maximum of 220 gene copies per litre at
any station. If one assumes the highest cell-specific rate measured for
this organism (~50 fmol N cell−1 d−1, see refs. 12,13) and assumes one
nitrogenase gene copy per cell, one would calculate an equivalent
rate of ~0.01 nmol N l−1 d−1, which is orders of magnitude too low
to account for the reported rate at the ice-edge station. This analy-
sis remains valid even when a recently described correction factor
is applied to account for underestimated per-cell rates resulting
from isotope dilution effects14. Variability in UCYN-A distributions
Questioning High Nitrogen Fixation Rate
Measurements in the Southern Ocean
Angelicque E. White 1 ✉ , Julie Granger2 and Kendra Turk-Kubo 3
arising from T. Shiozaki et al. Nature Geoscience https://doi.org/10.1038/s41561-020-00651-7 (2020)
NATURE GEOSCIENCE | VOL 15 | JANUARY 2022 | 29–30 | www.nature.com/naturegeoscience 29
Content courtesy of Springer Nature, terms of use apply. Rights reserved