84
Chapter 5: Nutrient Limitation: The Example of Iron
active, because it is quite certain that many other factors, not considered, are involved
in the complicated processes that lead to an algal bloom. In the Southern Ocean blooms
are largely restricted to meandering frontal zones and are possibly enhanced downstream
from shallow topography, as I have already discussed.
Then, the authors find it informative that both dust deposition and surface chlorophyll
have seasonal maxima in the Southern Ocean in midsummer. They did not, apparently,
consider that each of these cycles is an independent response to sun angle. The algal
response is obvious at high latitudes, but it is not so well known that the dust-carrying
winds from Patagonia are “monsoonal,” so that dust flies NW in winter and SE in
summer. Nevertheless, the Erickson et al. model has been influential.
The wide range of assumptions used for the input field of Fe at the sea surface
indicates the level of uncertainty that is still inherent in geochemical nutrient models.
To obtain this input field in models, extrapolation has been made from observations
of haze at sea (Duce and Tindale, 1991), from global vegetation cover data (Mahowald
et al., 1999), from vegetation, soil texture, and land surface modification data (Tegen and
Fung, 1994), and from the location of topographic depressions (Ginoux et al., 2001). If
Fe is 3.5% of deposited dust (the usual assumption), the estimates of Fe deposition on
the ocean ranges sixfold from 15 to 100 Tg y
−1 so the global pattern of deposition flux
cannot yet be narrowly constrained. Yet it is probable that only between 5 and 20
N
in the eastern Atlantic, and seasonally in the extreme NW Arabian Sea, is flux greater
than 1–5 g m
−2 yr
−1 . In the southern hemisphere, it seems to exceed 025 m
−2 yr
−1 only
patchily, in some small regions, and seasonally.
We remain uncertain even of the relative magnitude of the fluxes of labile iron across
the sea surface and across the nutricline, although this must be critical to any quantitative
understanding of Fe limitation in the ocean. For instance, the suggestion of de Baar
et al. (1995) that an aerosol flux of 30 mg m
−2 y
1 dominates over vertical flux across
the nutricline in the North Pacific is based on what they called an “order of magnitude
assessment” of Duce and Tindale’s classical map. Modeling of the atmospheric flux has
given equally diverse results: Fung et al. (2000) simulated a deposition rate almost twice
that estimated by the model of Mahowald et al. (1999) and almost four times what was
suggested by Duce and Tindale’s data (263 and 132 × 10
9 mol y
−1 , respectively).
Fung et al. remarked that uncertainty concerning aeolian flux to the photic zone is
as large as a factor of 5–10, with an order of magnitude uncertainty in determining
the soluble fraction. Despite this, they asserted that entrainment and upwelling deliver
only 07 × 10
9 mol Fe/y
−1 across the nutricline, as against 960 × 10
9 mol Fe/y
−1 across
the surface: the consequences of entrainment and upwelling flux of Fe are trivial for
these authors. On the other hand, Archer and Johnson (2000) compare aeolian flux from
the sources quoted earlier and conclude that 70–80% of global carbon export production
by phytoplankton can be supported by the “upwelling of iron in seawater rather than
by atmospheric deposition.” These authors go on to propose that “ocean recycling of Fe
appears to play a major role in determining the strength of the biological pump in the
ocean and the pCO 2 of the atmosphere.”
Three recent intermediate-complexity ecosystem models appear to support, inadvertently, the primacy of physical processes in determining the pattern of productivity in the
ocean. Each is coupled to an OGCM and each comprises several classes of phytoplankton with appropriate herbivore classes and sinking terms. The models are distinguished
principally by their nutrient assumptions. The first has three macronutrients with no Fe
input, the second by the same team (Gregg et al., 2003) is similar, but has Fe input from
the GOCART model of Ginoux et al. (2001), and the third (Moore et al., 2002) has a
full range of nutrients (NO 3 NH 4 SiO 3 PO 4 , and Fe). In this case, Fe input is from
the models of Tegen and Fung (1994) and Mahowald et al. (1999), with solubility set at
2%. A fourth model (Christian et al., 2002) is structured like that of Gregg et al. (2003),
Chapter 5: Nutrient Limitation: The Example of Iron
active, because it is quite certain that many other factors, not considered, are involved
in the complicated processes that lead to an algal bloom. In the Southern Ocean blooms
are largely restricted to meandering frontal zones and are possibly enhanced downstream
from shallow topography, as I have already discussed.
Then, the authors find it informative that both dust deposition and surface chlorophyll
have seasonal maxima in the Southern Ocean in midsummer. They did not, apparently,
consider that each of these cycles is an independent response to sun angle. The algal
response is obvious at high latitudes, but it is not so well known that the dust-carrying
winds from Patagonia are “monsoonal,” so that dust flies NW in winter and SE in
summer. Nevertheless, the Erickson et al. model has been influential.
The wide range of assumptions used for the input field of Fe at the sea surface
indicates the level of uncertainty that is still inherent in geochemical nutrient models.
To obtain this input field in models, extrapolation has been made from observations
of haze at sea (Duce and Tindale, 1991), from global vegetation cover data (Mahowald
et al., 1999), from vegetation, soil texture, and land surface modification data (Tegen and
Fung, 1994), and from the location of topographic depressions (Ginoux et al., 2001). If
Fe is 3.5% of deposited dust (the usual assumption), the estimates of Fe deposition on
the ocean ranges sixfold from 15 to 100 Tg y
−1 so the global pattern of deposition flux
cannot yet be narrowly constrained. Yet it is probable that only between 5 and 20
N
in the eastern Atlantic, and seasonally in the extreme NW Arabian Sea, is flux greater
than 1–5 g m
−2 yr
−1 . In the southern hemisphere, it seems to exceed 025 m
−2 yr
−1 only
patchily, in some small regions, and seasonally.
We remain uncertain even of the relative magnitude of the fluxes of labile iron across
the sea surface and across the nutricline, although this must be critical to any quantitative
understanding of Fe limitation in the ocean. For instance, the suggestion of de Baar
et al. (1995) that an aerosol flux of 30 mg m
−2 y
1 dominates over vertical flux across
the nutricline in the North Pacific is based on what they called an “order of magnitude
assessment” of Duce and Tindale’s classical map. Modeling of the atmospheric flux has
given equally diverse results: Fung et al. (2000) simulated a deposition rate almost twice
that estimated by the model of Mahowald et al. (1999) and almost four times what was
suggested by Duce and Tindale’s data (263 and 132 × 10
9 mol y
−1 , respectively).
Fung et al. remarked that uncertainty concerning aeolian flux to the photic zone is
as large as a factor of 5–10, with an order of magnitude uncertainty in determining
the soluble fraction. Despite this, they asserted that entrainment and upwelling deliver
only 07 × 10
9 mol Fe/y
−1 across the nutricline, as against 960 × 10
9 mol Fe/y
−1 across
the surface: the consequences of entrainment and upwelling flux of Fe are trivial for
these authors. On the other hand, Archer and Johnson (2000) compare aeolian flux from
the sources quoted earlier and conclude that 70–80% of global carbon export production
by phytoplankton can be supported by the “upwelling of iron in seawater rather than
by atmospheric deposition.” These authors go on to propose that “ocean recycling of Fe
appears to play a major role in determining the strength of the biological pump in the
ocean and the pCO 2 of the atmosphere.”
Three recent intermediate-complexity ecosystem models appear to support, inadvertently, the primacy of physical processes in determining the pattern of productivity in the
ocean. Each is coupled to an OGCM and each comprises several classes of phytoplankton with appropriate herbivore classes and sinking terms. The models are distinguished
principally by their nutrient assumptions. The first has three macronutrients with no Fe
input, the second by the same team (Gregg et al., 2003) is similar, but has Fe input from
the GOCART model of Ginoux et al. (2001), and the third (Moore et al., 2002) has a
full range of nutrients (NO 3 NH 4 SiO 3 PO 4 , and Fe). In this case, Fe input is from
the models of Tegen and Fung (1994) and Mahowald et al. (1999), with solubility set at
2%. A fourth model (Christian et al., 2002) is structured like that of Gregg et al. (2003),