Estimation of the Internal P Source and Sink in the Baltic Proper
We will apply Eq. (
5) to the situation in 1980 and 2005, respectively, because they represent years with a small and large extent of anoxic area (Fig.
3) and a decreasing trend in external P load (Fig.
1). We then get two equations from which we can estimate the magnitude of the difference
Intsink −
Intsource. The total P content, increased from about 550 000 tons in 1980 to about 700 000 tons in 2010 (Fig.
2), i.e., the long-term mean of
\( V\frac{{{\text{d}}\bar{c}}}{{{\text{d}}t}} \), can be estimated to about 5000 tons P year
−1.
Extsource is shown in Fig.
1. The export of P is partitioned into two terms, the first, barotropic, term is the product of the freshwater run-off
Q
f and the annual mean of the concentration
\( c_{1} \) in the surface layer, which we estimate to at most 0.5 mmol P m
−3 in 1980 and 20 % higher in 2005; c.f. the change of P content in the upper layer (Fig.
2) in this time interval. The canonical value of
Q
f equals 450 × 10
9 m
3 year
−1. This gives a contribution from the barotropic term to the annual export of at most 7000 and 9000 tons P year
−1 in 1980 and 2005, respectively. We neglect variations in export due to variations in
Q
f. The second, baroclinic, term depends on the product of the rate of inflow
Q
1 and the annual mean concentration difference
\( \left( {c_{1} - c_{0} } \right) \) between the surface water in the Baltic proper and the inflowing Kattegat water. This difference is probably close to zero, it may even be negative and if this happens, the baroclinic term turns into an import. The salinity balance of the Baltic proper shows that the long-term mean inflow is about equal to the rate of freshwater supply, i.e.,
Q
1 ≈
Q
f (e.g., Stigebrandt
2001). For the time period considered, we estimate the net export of P to the Kattegat to be in the interval 5000 to 10 000 tons P year
−1. This is in accordance to the estimated net flux, about 9500 tons P year
−1, obtained by Rasmussen and Gustafsson (
2003) based on a thorough analysis. The export in the 1970s was estimated to 5500 tons P year
−1 (Wulff and Stigebrandt
1989).
For 1980 we estimate
Extsource = 60 000,
\( V\frac{{{\text{d}}\bar{c}}}{{{\text{d}}t}} \) = 5000 and
Extsink = 7000, thus
$$ Intsink-Intsource = 48\,000\,{\text{tons}}\,{\text{P}}\,{\text{year}}^{ - 1} . $$
(7)
For 2005 we estimate
Extsource = 35 000,
\( V\frac{{{\text{d}}\bar{c}}}{{{\text{d}}t}} \) = 5000 and
Extsink = 9000, thus
$$ Intsink-Intsource = 21\,000\,{\text{tons}}\,{\text{P}}\,{\text{year}}^{ - 1} . $$
(8)
The model thus shows that Eq. (
5) is balanced if
Intsink −
Intsource decreases from 48 000 tons P year
−1 in 1980 to 21 000 tons P year
−1 in 2005, meaning that there is a net internal source in the Baltic Sea. This is the major result obtained from the model. However, to explain the changes observed in the Baltic Sea, we couple this result to the changing area of the anoxic bottoms in the Baltic proper. We study the case outlined in “
Materials and Methods”, where anoxic bottoms act as net P sources and the sink is distributed over the total area. We choose to focus on this case, since it gives a specific P flux from anoxic bottoms (
fs) that is commensurable with other independent estimates as shown below. Other cases are possible, e.g., only oxic bottoms act as sinks of P and no P is buried at anoxic bottoms; such a scenario is briefly discussed in “
A New P-Paradigm—Anoxic Bottoms as Temporal Net P Sources”. To determine
Intsource we use Eq. (
6) with
\( A_{\text{anox}} \) equal to 20 000 and 40 000 km
2, respectively, in 1980 and 2005 (Fig.
3, dashed line). For
Intsink we use Eq. (
2) with
c(2005) = 1.2·
c(1980), i.e., we assume that the winter surface concentration has increased by 20 % from 1980 to 2005 as described above. Using Eqs. (
2) and (
6) we rewrite Eqs. (
7) and (
8) and get the following equations for 1980 and 2005, respectively,
$$ c \cdot v \cdot A = fs \cdot 20\,000 + 48\,000 $$
(9)
$$ 1.2 \cdot c \cdot v \cdot A = fs \cdot 40\,000 + 21\,000. $$
(10)
Solving this equation system gives a 1980–2005 time averaged
fs = 2.3 g P m
−2 year
−1. This assumes that
fs has remained constant for the time period 1980–2005. The resulting
fs implies that
Intsource = 45 750 tons P year
−1 in 1980 and
Intsource = 91 500 tons P year
−1 in 2005 (from Eq.
6). For
Intsink we get from Eqs. (
7) and (
8) 93 750 tons P year
−1 in 1980 and 112 500 tons P year
−1 in 2005, respectively.
This model solution suggests that the internal source doubled in the period from 1980 to 2005 due to a doubling of the area covered with anoxic water. DIP released from deep anoxic bottoms is thus transported upwards by water circulation and, via production of POP in the surface layer, it may reach sinks at shallower areas. A similar mechanism, the P shuttle, does the same but is based on the redox-controlled Fe- and Mn-oxides’ ability to scavenge phosphate. This has been discussed and quantitatively estimated in the context of the Black Sea (Shaffer
1986) and recently applied to the Baltic proper pelagic redox-cline (Turnewitsch and Pohl
2010).
Intsink is an order of magnitude larger than
Extsink. Some of the internal sink may be accounted for by export from the Baltic proper to the Bothnian Sea. It was estimated to be about 3900 tons P year
−1 in the 1970s (Wulff and Stigebrandt
1989) and 5800 tons P year
−1 in the 1980s (Wulff et al.
2001). Also the external P load to the Bothnian Bay, ca 8100 tons P year
−1 in the 1980s (Wulff et al.
2001), should remain as sinks in this sea. These estimates show that about 85 % of our estimated internal sink in the Baltic Sea should be located within the Baltic proper, as sediment burial.
Application of our P model gives
fs = 2.3 g P m
−2 year
−1 from anoxic bottoms and this can be compared to other estimates based on large-scale budgets. Conley et al. (
2002) found that year-to-year increases of the phosphorus content in hypoxic water could be explained by sediment releases of on average 2 g P m
−2 year
−1 with a maximum release of 5 g P m
−2 year
−1 from hypoxic and anoxic bottoms. Similar fluxes of 2 g P m
−2 year
−1 from anoxic bottoms in the Eastern Gotland Basin (EGB) using the budget method described in
Electronic Supplementary Material were estimated in Gustafsson and Stigebrandt (
2007). The benthic DIP flux estimated with our P model can be validated by reliable direct measurements. By measurements in situ using benthic landers, DIP fluxes in the EGB of 4.2 ± 2.4 g P m
−2 year
−1 from bottoms overlain by anoxic water have been reported (Viktorsson et al.
2013). Estimates using hydrographical observations in the stagnant Bornholm Basin presented in “
Phosphorus Flux from Bottom Sediments in the Bornholm Basin” show 3–5 times larger benthic DIP fluxes under anoxic than under oxic conditions. The in situ flux method and the basin budget method include both a possible source from anoxic bottoms and the reflux due to consumption of fresh organic matter. Therefore, they should give a larger flux than the flux (
fs) from the model.
Phosphorus Flux from Bottom Sediments in the Bornholm Basin
Using a budget method the oxidation rate of organic carbon, 28 g C m
−2 year
−1, was calculated by Stigebrandt and Kalén (
2013) for the Bornholm Basin (BB) for the 1980s and later. Assuming that the organic matter is composed according to the so-called Redfield ratio, 0.7 g P m
−2 year
−1 should be remineralized to DIP. Consequently, the downward flux of P bound to POP is at least 0.7 g P m
−2 year
−1. According to these authors, oxygen consumption in the BB doubled from the 1960s and 1970s to the 1980s, meaning that NP in the 1960s was half of that in the 1980s. Please note that the budget method used by Stigebrandt and Kalén (
2013) is insensitive to whether organic matter is entering directly from the surface layer or via lateral transport in the benthic boundary layers.
To calculate
FS, the specific DIP flux from the sediments below 75 m, in the BB, a budget method (described in
Electronic Supplementary Material) has been applied to the data set of total phosphorus from station BY5 (data available at
www.smhi.se). The results for five decades, starting with the 1960s, are given in Table
1 where fluxes are presented separately for oxic and anoxic conditions in the water. Anoxic conditions culminated in the 1980s when the basin was anoxic about 20 % of the time. However,
FS during anoxic conditions culminated in the 1990s when it reached the rate 8.6 g P m
−2 year
−1. The uncertainty of
FS due to the uncertainty of the value of
the vertical diffusivity is about ±20 % (see Table
1).
Table 1
Flux FS of DIP (g P m−2 year−1) from the bottom sediments beneath 75 m depth in the Bornholm Basin during oxic (FS oxic) and anoxic (FS anoxic) conditions during five decades. The percentage of the time the basin water has been oxic and anoxic is given as well as the number of estimates (No. estimates). The weighted averages (FS aver) for the basin accounts for the percentage of time that the basin water is oxic and anoxic, respectively
1960s | 0.8 ± 0.2 | 100 | 25 | – | 0 | 0 | 0.8 |
1970s | 1.0 ± 0.2 | 96.7 | 25 | 3.0 ± 0.4 | 3.3 | 1 | 1.1 |
1980s | 1.1 ± 0.3 | 79.7 | 16 | 3.8 ± 0.8 | 20.3 | 4 | 1.6 |
1990s | 1.7 ± 0.2 | 88.1 | 43 | 8.6 ± 1.6 | 11.9 | 11 | 2.5 |
2000s | 1.5 ± 0.3 | 86.1 | 60 | 6.5 ± 0.8 | 13.9 | 13 | 2.2 |
FS was typically 3–5 times greater during anoxic than during oxic conditions (Table
1).
FS was larger than the downward flux of POP, as calculated from oxygen consumption above, both during oxic and anoxic conditions. It was twice the estimated downward flux of POP during oxic periods. In our analysis, conditions are defined as anoxic when the water is anoxic. However, DIP fluxes from the sediment should be more sensitive to whether or not the sediment–water interface is oxidized. This should prolong anoxic conditions since it takes time to oxidize the interface. This may explain why
FS was larger than the downward POP flux also during oxic conditions in the water column.
FS was more than 10 times larger than the downward flux of POP during anoxic conditions. This means that most of the DIP released under anoxic conditions must come from a storage that probably accumulated during earlier oxic periods. It is known that even the deepest parts of the BB were oxic and inhabited by, e.g., the long-lived bivalve
Macoma calcarea in the first half of the twentieth century (e.g., Gerlach
1994). It is not known when this oxic period started but it ended between 1948 and 1956 as suggested by Gerlach (
1994).
The last column in Table
1 gives the weighted decadal average of DIP flux from the sediments below 75 m depth (
FS aver). Using this, the DIP flux in the 1960s was about 0.8 g P m
−2 year
−1. With
A(75) = 5000 km
2, the annual upward flux of DIP from the bottoms below 75 m was about 4000 tons P year
−1 for this decade. It then increased and was about 11 000 tons P year
−1 in the 2000s. The estimated supply of P by settling organic matter (POP) is only about 1800 and 3500 tons P year
−1, respectively, for the two decades, as estimated above from the oxygen consumption. Thus, the deeper parts of BB appear to have acted as an internal source of about 2200 and 7500 tons P year
−1 for the two decades, respectively. The accumulated net loss of P (upward DIP minus downward POP) from the deep bottoms in BB since 1960 should be about 50 g P m
−2 (i.e., on average 1 g P m
−2 year
−1). This is less than the value of
fs from the model for the Baltic proper, which is expected because, after all, the BB has been oxic most of the time.