Deep-Water Oxygen Conditions in the Bothnian Sea

Deep-water oxygen conditions in the Bothnian Sea

Mika Raateoja

Finnish Environment Institute (SYKE), Marine Research Centre, P.O. Box 140, FI-00251 Helsinki

Received 17 July 2011, final version received 25 Sep. 2012, accepted 24 Apr. 2012

Raateoja, M. 2013: Deep-water oxygen conditions in the Bothnian Sea. Boreal Env. Res. 18: 235–249.

The hydrography of the deep waters of the Bothnian Sea (BS) underwent considerable changes during the most recent two decades. Most importantly, slightly worsened deep-

water oxygen (O2) conditions followed a gradual increase in the water-column density gra-

dient. Up to the present time, the lowest measured O2 concentrations fell within the range 3.5–4.5 ml l–1, 1.4 ml l–1 being used in this work as the upper limit for hypoxia. I investi- gated the long-term deep-water characteristics of the BS to determine how possible it is for an hypoxic event to occur there in the near future. It appears that there is only a remote probability of this, unless profound changes take place in the hydrographic regime of the BS and/or in the anthropogenic load to the BS. The key factor behind this conclusion is the hydrodynamic regime of the Åland Sea and the ridge formations around it. This area acts

as a buffering/filtration mechanism that prevents the saline and O2-poor deep waters of the Northern Baltic Proper from entering the BS without considerable mixing and dilution. A discussion is presented of the consequences of this conclusion, taking into account the past adverse development in the Gulf of Finland, with its internal loading of phosphorus and its

accelerated eutrophication. With respect to the Finnish coastal area of the BS, the stable O2 conditions and the favourable hydrographic setup in the near-coastal area suggest that any

O2 problems will most probably be concentrated in the inner-archipelago areas.

Introduction Sea. The Baltic Sea has an intrinsic tendency

to experience O2 problems due to its natural The spreading of hypoxia in coastal and estua- vertical salinity stratification, and hence, den- rine marine environments is currently a ubiqui- sity gradients that effectively restrict the vertical tous problem. Coastal systems that experience transport of O2. These salinity anomaly layers, episodic hypoxic events or permanent hypoxia called haloclines, have played a pivotal role in have greatly increased in numbers (Selman et the spreading of a hypoxic environment in the al. 2008). The conspicuous factor behind this Baltic Sea; indeed, O2 problems have persisted adverse development is eutrophication, which there for as long as the present geographical boosts the flow of organic matter to the ben- coverage of the halocline system has prevailed, thic system, ultimately degrading the oxygen that is, from the commencement of the Litto-

(O2) conditions there. This development has also rina Sea stage at approximately eight millennia taken place in the Baltic Sea. BP. (Zillen et al. 2008). From this perspective, However, eutrophication is not solely to anthropogenic eutrophication has had little time blame for the current poor O2 status of the Baltic to leave its fingerprints in the O2 regime of the Editor in charge of this article: Kai Myrberg 236 Raateoja • Boreal Env. Res. vol. 18

of Bothnia is not serious, the increasing trends Northern Quark in the nutrient levels should be seen as warning signs for the future”. A closer look into the BS US2 US7 system is indeed topical.

US5B Eutrophication is a process that, if given a chance, follows the cause-and-effect cascade MS11 of increased allochthonous load of nutrients  increased autochthonous load of organic matter SR5 SR9 61°N, 23°E  anoxia  increased autochthonous load of nutrients (the internal loading of phosphorus). Southern Quark There exists a strong feedback path from the

F64 end-product, the internally-loaded phosphorus (P), adding the autochthonous component to Southern Åland Sill LL7 eutrophication. This feedback mechanism has the potential to strongly accelerate the eutrophication LL19 process. I wanted to clarify whether this cascade, which has already commenced and is still on- going in the GOF, could be a potentially realis- Fig. 1. Map of the study area. Stations SR5, US2, tic scheme for the BS. In this article, I present and US5B were chosen as the main data sources for a long-term hydrographic dataset collected at the study. Stations SR9, MS11, and US7 were used representative monitoring stations in the BS. I to describe the oxygen conditions in the Finnish near- address the following questions: (1) how have coastal zone of the BS. The cascade of stations SR5, F64, and LL19 reflected the hydrographic variation the near-bottom O2 conditions in the BS recently between the NBP and the BS. The station LL7 in the developed? (2) what are the external factors that

GOF served as the basis for comparison for the deep control the O2 regime of the BS? and (3) what can water nutrient conditions. we expect to happen in the near future?

Baltic Sea. Today, hypoxic environments con- The Bothnian Sea sistently cover large offshore areas in the Baltic Proper, while the western parts of the Gulf of The BS is the southern basin of the Gulf of Finland (GOF) can be classified as seasonally Bothnia, which is the semi-enclosed northern hypoxic (HELCOM 2009b); furthermore, the extension of the Baltic Sea. In the Baltic Sea, the hypoxic environment has recently been spread- BS is the largest basin after the Gotland Basin, ing in the coastal areas (Conley et al. 2011). with an area of about one-sixth of the Baltic Sea The offshore Bothnian Sea (BS), on the other area and a water volume of about one-fifth of hand, has not been further eutrophied in terms the Baltic Sea volume (Leppäranta and Myrberg of dissolved inorganic nutrients since the early 2009). The average depth of the BS is 66 m as 1990s (Fleming-Lehtinen et al. 2008, HELCOM compared with 54 m for the Baltic Sea; the BS 2009b). Even so, alarming signals have emerged contains a central plane with large areas deeper during this period, e.g., the chlorophyll a levels than 100, or even 200 m (Fig. 1). and cyanobacterial biomasses have steadily The southern Åland Sill, the southern Quark, increased there since the mid-1990s (Fleming- and the shallow and scattered Archipelago Sea Lehtinen et al. 2008, Jaanus et al. 2011). The isolate the BS effectively from the Northern Finnish coastal area of the BS suffers from the Baltic Proper (NBP). Consequently, the hydro- allochthonous nutrient load, and the inner coastal graphical forcing of the NBP on the BS is area experiences episodic hypoxic events (Lun- limited, albeit constantly present. The northern dberg et al. 2009). Lundberg and co-workers Quark, which topographically separates the BS stated that “… even if eutrophication in the Gulf from the Bothnian Bay, does not restrict the Boreal Env. Res. vol. 18 • Deep-water oxygen conditions in the Bothnian Sea 237 southerly flow of fresher water to the BS. This Material and methods flow and the direct river runoff into the BS together create a pronounced fresh water impact Stations on the BS. Together these topographic features result in hydrological properties for the BS that The main stations used in this study (SR5, US2, are markedly different from those in the NBP. and US5B) are located in the central plain of the The pronounced river runoff into the Gulf of BS at water depths > 100 m (Fig. 1). These sta- Bothnia leads to a short water residence time on tions are situated in the sediment-accumulation the Baltic Sea scale, that is, 5 to 6.5 years (Myr- areas that are the first to experience O2 problems. berg and Andrejev 2006). In this deep area, a consistent albeit faint halo- The BS freezes over, at least partly, every cline represents a quasi-stable hydrographic bar- winter. The probability of freezing is highest in rier for the vertical flow of O2 to deeper waters, the Finnish coastal area and the areas next to the emphasizing the role of the O2 supply via hori- northern Quark that freeze every winter, becom- zontal advection (Table 1). US2 is located in the ing smaller in the southern half of the offshore northwestern part of the BS, and is hence subject area; this remains ice-free during mild winters, to the impact of the southward surface current of i.e., about one-third of the winters (Feistel et al. less saline water from the Bothnian Bay. SR5 is 2008). located in the southern part of the central plain, Every summer, a thermocline develops in the and its deep-water hydrographical characteristics BS and this acts as a barrier for vertical mixing are strongly influenced by the hydrographical at a depth ranging from 10 to 20 m depending on forcing of the Åland Sea (ÅS). the year and the season. It separates the surface The near-coastal stations of the BS (SR9, water mass with temperatures of 16–18 °C at its MS11, and US7) are located in the Finnish outer maximum from the deep waters at 2–5 °C. coastal area of the BS. The water depth is < 50 m, The BS has a cyclonic surface water circula- and the virtually non-existing vertical density tion, which leads to a less saline surface layer on gradient in winter indicates that the halocline its deeper western side, and a more saline surface does not extend to these shallow areas (Table 1). layer on its shallower eastern side. This general The station representative for the ÅS, F64, is picture is however strongly subject to wind forc- located in the northern part of the ÅS in a trench ing, and is thus intermittent (Håkansson et al. that extends over most of the basin. The sea area 1996). The surface salinity of the BS ranges is subject to efficient vertical mixing and no from about 4.5 in the north to about 5.5 PSU in hypoxia is observed there, even in deep waters the south. The input of fresher water from the (down to about 300 m). Bothnian Bay produces marked variation in this The station representative for the NBP, LL19, typical value, especially in the northwestern part is situated in the midst of the northern Gotland of the BS. Salinity increases quite steadily with Basin. With a water depth of > 150 m, the sea depth, somewhat faster in the halocline located floor there is typically anoxic. typically at a depth of 50–75 m. However, no The station representative for the GOF, LL7, distinct halocline like that in the NBP is found. is located in the western part of the gulf in an The salinity near the sea floor is typically about area that is characterized by episodic and unpre- 1 PSU higher than that at the surface. This faint dictable anoxic events due to the hydrographical halocline does not create as strong a water- forcing of the NBP; hypoxia is met with there column density gradient as that in the NBP, for frequently. example, and consequently the potential of the gradient to restrict vertical mixing varies consid- erably (Kullenberg 1981). The stratification in Data sources the BS is thus governed to a much greater extent by temperature variations than is typical for the The marine dataset was compiled within the southern basins of the Baltic Sea. Baltic monitoring programme (BMP), and later, 238 Raateoja • Boreal Env. Res. vol. 18 2 5.9 5.3 5.3 8.6 8.8 8.7 6.5 2.5 2000s ) –1 (ml l 6.7 5.8 5.7 8.4 8.3 8.2 6.8 3.3 all data M ean near-bottom O

t 0.72 0.94 0.62 0.10 0.13 0.10 1.04 3.23 a noxic 1.99 ) –3 (Jan– M ar) winter data (kg m M ean ∆ σ

1.40 1.12 1.18 0.47 0.54 0.31 1.67 4.09 2.80 all data concentrations. 2 41 36 25 (m) 209 130 222 286 169 102 Depth

2 3 6 6 1 ), and near-bottom O t 10 19 12 18 ∆ σ no visits number of years with 10 43 17 37 18 43 28 31 12 scheme < 3 visits S ampling number of years with

49 50 46 38 19 46 49 47 50 years of monitoring

72 26 14 48 94 245 132 109 359 of visits

1962–2011 1966–2011 1963–2011 1965–2011 1962–2011 Data range n umber 1962–2010 1972–2009 1990–2008 1965–2010 U S 2 U S 5B U S 7 F64 tations included in the study, sampling scheme, water-column density ranges ( 1 . S tations included in the study, Table S tation Deep stations sr 5

Near-coastal stations sr 9

ms 11

Stations south of the BS ll 19 Station in the GOF ll 7 Boreal Env. Res. vol. 18 • Deep-water oxygen conditions in the Bothnian Sea 239

–1 the HELCOM/COMBINE programme, carried Since 1962, the O2 concentration (ml l ) out by the Finnish Institute of Marine Research was analyzed employing the Winkler technique (1962–2008) and later by the Marine Research according to Grasshoff et al. (1999) using either Centre of the Finnish Environment Institute visual, or later, potentiometric endpoint recogni-

(2009–2010). The sampling scheme at the station. Since 1999, the CTD-based O2 was meas- tions is shown in Table 1. Typically, all existing ured by Clark-type electrodes of various CTD data sources were used: numerical data collected systems that were calibrated by the manufac- with CTD sensors, and analytical data deter- turer on a constant basis, and whose output was mined both from CTD’s rosette samples and from checked against wet O2 analytics at every sta- discrete water samples. The riverine dataset was tion. The upper limit for hypoxia was defined compiled within the HELCOM/PLC programme. according to Conley et al. (2009) who sug- –1 With respect to the water column, this study gested a value of 2 mg O2 l (equivalent to ≈ 1.4 –1 concentrates on the surface layer (above any ml O2 l ). The O2 saturation (%) was calculated density gradient) and the deep layer (below any from the O2 concentration and temperature. density gradient). The combined data from the The concentration of the dissolved inorganic –1 depths of 0 and 5 m describe the surface layer, phosphorus (PO4-P, µmol l ) was determined which is denoted “SURFACE”. The deep layer is colorimetrically, first using manual methods described by the data collected as follows: SR5 ≥ employing the photometric determination of 100 m, US2 ≥ 175 m, US5B ≥ 175 m up to 1988 the substance (Grasshoff et al. 1999), and later and ≥ 200 m since 1989, SR9 ≥ 35 m, MS11 ≥ using the FIA approach with a SKALAR 5101 30 m, US7 ≥ 20 m, F64 ≥ 275 m, LL19 ≥ 150 segmented flow analyzer (1993 to 1999) and a m, and LL7 ≥ 70 m up to 2002 and ≥ 90 m since Lachat QuickChem 8000 flow injection analyzer 2003. The deep layer is denoted “DEEP”. (1999 to 2011). The outputs of the instruments were internally validated against each other. Currently, the measurement accuracies

Parameters of PO4-P, salinity and dissolved O2 are ±22%, ±0.3% and ±7%, respectively. For salinity, the practical salinity scale (PSS78) was employed, and the practical salinity unit (PSU) was used throughout the study (UNESCO Quality assurance 1980). Since 1962, the salinity determined from discrete water samples was measured with For salinity and O2, the chemical foundations various Autosal bench salinometers calibrated of the measuring techniques as well as the basic against IAPSO standard seawater (Grasshoff et technical principles of the instruments did not al. 1999). Since 1978, the CTD-based salinity changed during the data collection period pre- was measured with conductivity cells of various sented in this article. For PO4-P, the change from CTD systems whose output was checked against the manual procedure to the FIA technique did not wet analytics on a constant basis. change the optical setup, as the chemical founda- Since 1962, the temperature (°C) at the sam- tion remained the same. Thus, with regard to these pling depth of the discrete water samples was parameters, the only plausible source of error measured with thermometers installed on the related to quality control stems from the inevita- sampling bottles. Since 1978, the CTD-based ble changes in the instrumentation. This source temperature was measured with the temperature of error has been minimized with the instrument probes of various CTD systems that were cali- validation and inter-comparison procedures. brated by the manufacturer on a constant basis. All the methods above employed by the In the shallow waters of the Baltic Sea, these two Finnish Environment Institute and the Finnish parameters determine the density σt (= density – Meteorological Institute are currently internally 1000 kg m–3) (Millero and Poisson 1981). The accredited by the Finnish Accreditation Service water-column density range was defined as ∆σt = (FINAS) following the SFS-EN ISO/IEC 17025

σt(DEEP) – σt(SURFACE). standard. 240 Raateoja • Boreal Env. Res. vol. 18

SR5 US2 US5B 7.5

7.0 )

6.5

Salinity (PSU 6.0

5.5 6 ) 4

2 Temperature(°C

0 6.0 ) –3 5.5 , kg m t

5.0 Density ( σ

4.5 100 )

60 Oxygen saturation (%

40 1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010 Fig. 2. Trends in the hydrographic characteristics in the DEEP at the offshore stations of the BS. Both the available CTD data and the results from the discrete samples are included. The datasets are loess-smoothed when appropri- ate and the resulting fits are presented with solid lines.

Results reflects the DEEP density because, in the BS, density variation is almost solely governed by DEEP hydrography in the BS salinity variation, except in the surface layer in summer (Table 2A). The DEEP temperature does The DEEP salinity in the BS underwent promi- not present such a distinct fluctuating pattern but nent changes during the past five decades (Fig. 2). showed a consistent, albeit faint, increase during Various phases emerged: (i) a phase of increasing the studied period. Any temporal trend in the salinity in the 1960s and 1970s ending up with DEEP O2 saturation is somewhat obscured by the the peak in the late 1970s, (ii) a phase of decreas- intrinsic seasonal variation linked to O2 dynam- ing salinity throughout the 1980s ending up with ics. Nevertheless, on the large scale, the DEEP O2 the low-point in the early 1990s, and (iii) a phase saturation decreased whenever the DEEP density of increasing salinity ever since. This pattern also increased. Boreal Env. Res. vol. 18 • Deep-water oxygen conditions in the Bothnian Sea 241 L inear dependences ( F , n p ) ( F , n p ) emperature 2 2 0.001 (0.09, 148, 0.76) 0.011 (1.9, 172, 0.17) 0.011 0.001) < 179, (12.8, 0.067 r r T 0.677 (522, 250, < 0.001) F64 – 0.583 (41.9, 31, < 0.001) L inear regressions describing the CT D probes and/or rosette samples, concentration reaches the upper limit of limit upper the reaches concentration 2 O n addition to being a consequence of natural variation natural of consequence a being to addition I n M aterial and methods. ( C ) CT D are the data from the ( F , n p ) ( F , n p ) ( F , n p ) 2 2 2 0.998 (59274, 149, < 0.001) 0.999 (182826, 174, < 0.001) 0.987 (13491, 179, < 0.001) 0.175 (38.3, 176, < 0.001) 0.177 (9.80, 41, < 0.01) 0.353 (4.27, 6, 0.094) r 0.413 (173, 250, < 0.001) r 0.420 (23.8, 34, < 0.001) S alinity SR 5 0.854 (187, 33, < 0.001) r arget year is the extrapolated year when the when year extrapolated the is year arget T

ime ime T T 2062 2082 2071 Jul– S ep Jul– S ep Jan– M ar Jan– M ar Jan– M ar Jan– M ar arget year T see the exact depth distributions in trends. 2 A nnual means are used since the early 1960s. ACE and D EE P a ≈ 0 in temporal O ) on water temperature and salinity at the offshore stations in the B S since the early 1960s. ( B ) t LL 19. S U R F

) as a function of time from 1990 to 2011. 2011. to 1990 from time of function a as ) Data Data –1 conc. = a ¥ year + b ) 2 CT D, D IS CT D, D IS CT D, D IS CT D, D IS CT D, D IS CT D, D IS SR 5, F64, and ) at t values resulted from 2 r conc. = –0.0714 ¥ year + 149 ¥ year + 87.0 conc. = –0.0411 conc. = –0.0359 ¥ year + 75.8 2 2 2 O O E quation ( O O nly data collected from the discrete samples taken ≤ 5 m above the seafloor were included. were seafloor the above m 5 ≤ taken samples discrete the from collected data O nly ). concentration (ml l (ml concentration –1 2 O L inear dependence of water density ( σ

ACE ACE concentrations, the low 2 D EE P D EE P F64 s U R F between the D EE P densities ( σ c ase 2 c ase 3 D IS are the data from discrete water samples, for the in decrease ll 19 Table 2 . A ) ( Table A s U R F B C c ase 1 hypoxia (≈ 1.4 ml l ml 1.4 (≈ hypoxia in O 242 Raateoja • Boreal Env. Res. vol. 18

SR5 US2 US5B 7.5

7.0 ) 6.5

6.0 Salinity (PSU 5.5

5.0 5

4 ) 3

1 Temperature(°C

–1 6.0

) 5.5 –3

5.0 , kg m t

4.5

Density ( σ 4.0 Deep Surface 3.5 1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 2020 Fig. 3. Trends in winter (Jan–Mar) hydrographic characteristics in the SURFACE and in the DEEP at the offshore stations of the BS. Both the available CTD-data and the results from the discrete samples are included. The datasets are loess-smoothed where applicable and the resulting fits are presented with solid lines.

Hydrographic variation between DEEP annual variation in the SURFACE — caused and SURFACE in the BS mainly by the severity of the early winter before the time of the station visits but also by the vari- The comparison between the temporal salinity able timing of the station visits — makes it dif- patterns in the DEEP and the SURFACE makes ficult to detect any definite temporal patterns. it obvious that the water-column salinity gradient has became larger since the low-point of the

DEEP salinity in the early 1990s (Fig. 3). The Near-bottom O2 in the BS dataset is not long enough to reveal any possible long-term fluctuations. In the shorter time inter- During the period since the low-point of the val, however: (i) the gradient in the 1970s was DEEP salinity in the early 1990s, both the near- about as large as that in the 2000s, (ii) the gradi- bottom density and the near-bottom O2 saturation ent became smaller from the late 1970s up to the contain fluctuations superimposed on the general low-point, mainly because of the decrease in the increasing and decreasing trends, respectively DEEP salinity, and (iii) the gradient has gradu- (Fig. 4A). Here, the density variation explains ally increased since the low-point, accompany- much of the observed variation in the O2 satura- ing the increasing DEEP salinity. The SURFACE tion. If we postpone the density data in time by salinity seems thus to have followed the DEEP two years, the density pattern closely resembles salinity pattern only remotely and with a time a mirror image of the O2 saturation pattern; i.e., lag. Again, the pattern above also describes the whenever the density increased, the O2 condi- behaviour of the SURFACE and DEEP density. tions worsened with about a two-year time lag, With respect to temperature, the strong inter- and vice versa. Boreal Env. Res. vol. 18 • Deep-water oxygen conditions in the Bothnian Sea 243

1.1 1.5 A Normalized y O

1.0 1.0 2 saturation Normalized densit

Density

O2 saturation 0.9 0.5 9 B ) –1 l 7

case 1 concentration (ml 5

2 case 2 O

case 3

1990 1995 2000 2005 2010 2015

–3 Fig. 4. (A) Trends in the near-bottom O2 saturation (%) and density (σt, kg m , data postponed in time by two years, i.e., the measuring time of σt was converted to equal the original time plus two years) at the offshore stations of the

BS. The O2 saturation and density datasets were normalized to unity. (B) Trends in the near-bottom O2 concentration at the offshore stations of the BS. In order to hypothesize the possible point of time when the upper limit for –1 hypoxia (≈ 1.4 ml O2 l ) would be reached, linear regression fits (solid line) with 95% confidence intervals (dashed and dotted lines) were included as follows: Case1 = the entire data estimating the emergence of hypoxia on a frequent basis, Case 2 = the lowest quartile data estimating the emergence of hypoxia on an occasional basis, and Case 3 = the lowest values picked visually estimating the emergence of hypoxia for the first time ever. For both of the figures, only the data from the discrete samples taken ≤ 5 m above the seafloor were included; subsequently, the data from all three stations were pooled.

The near-bottom O2 concentration in the BS the emergence of hypoxia on a frequent basis. has gradually decreased since the low-point of This seemingly impossible result indicates that, the DEEP salinity in the early 1990s (Fig. 4B). while the overall near-bottom O2 conditions have

During the studied period, the lowest O2 concen- slowly worsened, the lowest recorded values of —1 trations fell within the range of 4 to 5 ml l , with O2 concentration have decreased at a slower episodic events values of 3.5 to 4.5 ml l–1 being rate than the values in general, thus obscuring recorded. The average value for the entire dataset the linear extrapolations that reach out into the (Case 1) decreased from 6.6 at the low-point to future. 5.1 ml l–1 at the end of the study period. Regard- less of the approach used when choosing the dataset for the linear extrapolation (Cases 1–3), Impact of the NBP on the BS the predicted point in time for the emergence of hypoxia is pin-pointed as occurring more than As the NBP waters drift into the ÅS and further fifty years in the future (Table 2C). Most inter- enter the BS as a deep current, the water density estingly, the Case 3 estimate for the emergence gradually decreases through mixing processes. of hypoxia for the first time ever, produced a The density of the surface layer of the NBP is predicted point in time that was a decade later much closer to the density of the deep layer of than that produced by the Case 1 estimate for the ÅS than to the density of the deep layer of 244 Raateoja • Boreal Env. Res. vol. 18

0 10 A B

50 9

100 ) –3

8 , kg m 150 t Depth (m) 6 200 Density ( σ

SR5 250 5 F64 LL19 300 4 4 5 6 7 8 9 1960 1970 1980 1990 2000 2010 Density (σ , kg m–3) t LL19 surface F64 SR5 LL19

Fig. 5. The water-column density (σt) for the stations SR5, F64, and LL19. (A) The vertical pattern. The data were averaged over the years 1991–2010 (following the low-point of density), and the months Jan–Mar. (B) The temporal pattern in the DEEP at the stations, as well as in the SURFACE of LL19. Annual means were used. The SURFACE data for LL19 was available only since 1992. The DEEP values of LL19 give an indication of the potential density gradient occurring in the deep layer of the BS if the water entering the ÅS were to originate from below the halocline of the NBP. the BS (Fig. 5). The NBP water thus flows into in a fortunate position in that it is not so directly the ÅS relatively unchanged, and most of the connected to the NBP like the GOF is. mixing occurs in the ÅS. The close temporal resemblance of the SURFACE density of the NBP and the DEEP densities of the ÅS and the Hydrodynamic cascade of NPB–ÅS–BS BS underlines the close hydrodynamic connection between the basins (Fig. 5B and Table 2B). The hydrodynamic forcing of the NPB on the deep waters of the BS is mediated, and somewhat obscured, by the ÅS and the ridge forma- Discussion tions surrounding it to the north and south. The water flow from the NBP to the BS is most

External factors govern the O2 dynamics probably continuous, with the main mechanism of the BS being the winter cooling of the NBP water, its subsequent sinking over the Southern Åland Sill, The environmental status of the BS has received mixing in the ÅS, flowing over the Understen- fairly little public attention, mostly because the Märket sill of the southern Quark to eventually BS has not experienced such drastic environ- form the deep BS water mass (Marmefelt and mental problems like those that have emerged Omstedt 1993, Leppäranta and Myrberg 2009). elsewhere, e.g., in the GOF. With regard to O2 The first key aspect here is that the sill depth of dynamics, the problems in the GOF originate the southern Åland Sill between the NPB and the partly from the catchment (increased organic ÅS is 70 m (Marmefelt and Omstedt 1993), i.e., load, both allochthonous and autochthonous, either above or at the typical halocline depth of consuming more O2) but stem partly from the the NBP. The main bulk of the water entering the direct connection with the NBP (incoming highly ÅS, therefore, originates from above the halo- saline waters poor in O2). Here, I describe those cline of the NBP and not from the deep waters, hydrographical and/or topographical features which are highly saline and poor in O2 (Kullen- that have helped the BS to maintain the favour- berg 1981, Leppäranta and Myrberg 2009). This able O2 conditions in its deep layers. Those fea- aspect in itself has a drastic influence on the BS tures are external and, in this respect, the BS is deep water characteristics. Boreal Env. Res. vol. 18 • Deep-water oxygen conditions in the Bothnian Sea 245

The Southern Åland Sill largely determines Long-term changes in the BS salinity the characteristics of the waters flowing into the BS. Within this framework, the subsequent After its gradual increase spanning several dec- mixing occurring in the ÅS alters the character- ades and reaching its highest observed level of istics of the waters ultimately entering the BS. the last century in the late 1970s (Fonselius and The second key aspect here is that the inflow to Valderrama 2003), the DEEP salinity in the BS the BS takes place as a deep current, due to the took a different path and decreased throughout strong riverine-induced southward surface cur- the 1980s. A significantly decreased deep water rent through the ÅS. As the density difference inflow from the NBP (ICES 2008) as well as between the deep ÅS water and the water at the a sudden and distinctive increase in the river sill depth of the southern Quark is still signifi- inflow into the Gulf of Bothnia (Fig. 6) in the cant (Fig. 5A), the deep waters have to get less early 1980s probably had a synergistic effect that dense, i.e., less saline, to be able to cross the sill. enabled this alteration. Naturally, the change in Indeed, the deep water salinity changes at SR5 precipitation did not concern only the catchment have been reported to considerable resemble of the Gulf of Bothnia: in the same period, river those occurring at the sill depth of the Under- inflows increased in the Baltic Sea area in general sten-Märket trench (about 90 m) in the northern (Winsor et al. 2001). The progressively decreas- ÅS (Hietala et al. 2007). ing salinity of the BS was, along with increas- Taken together, the northward water flow ing temperature, the main driver for the regime over the ÅS can be seen as a two-step filtra- shift taking place in the BS in 1988–1989 (ICES tion mechanism: first, a clear cut-off mechanism 2008). This shift has been proposed to have been where the densest waters are prevented from driven by climate change (ICES 2008): the global entering the ÅS, and then a mixing/dilution warming period started in around 1980 (Lehmann mechanism which further shapes the water char- et al. 2011). It has also been explained by clima- acteristics before the water’s actual entry into the tological fluctuations: the NAO index increased BS. This mechanism is not expected to change to its highest positive observed level of the last unless there occurs an alteration in the water century from 1988 onwards (Kraberg et al. 2011, budget of the Baltic Sea that leads to a permanent Lehmann et al. 2011). Interestingly, the current rise in the halocline in the NBP. If this scenario dataset could not clearly reveal the increase in were ever to occur, the deep water characteristics temperature that evidently had taken place in the of the ÅS and BS would be reshaped towards a Baltic Proper (Alheit et al. 2005). The Baltic Sea state that would to some extent resemble the cur- renews most of its heat annually (Stigebrandt rent state of the NBP below the halocline, that is, and Gustafsson 2003), which is a challenge for more saline and poorer in O2. Alterations on this the conventional monitoring approach. This scale are only caused by widespread climatologi- approach may not be able to reveal trends in tem- cal changes over the northern hemisphere. The perature, which is a less conservative parameter scenarios linked to the on-going climate change than, e.g, salinity with a corresponding annual predict that, in the future, wintertime precipita- turnover proportion of only 3%. tion, and hence, river runoff, will increase in The trend in the DEEP salinity in the BS again quantity, and the inflow of saline water from took a different path in the early 1990s, and has the North Sea will decrease in quantity (BACC increased ever since. As the remnants of the major 2008, ICES 2008). Both of these tendencies Baltic inflows can be traced all the way to the work against the rise of the halocline in the NBP. Gulf of Bothnia (Fonselius and Valderrama 2003), To conclude, even though the salinity of the deep it is tempting to speculate that the upward step in inflow entering the BS has been increasing since the DEEP salinity in 1995–1996 was ultimately the low-point of the DEEP salinity in the early driven by the major Baltic inflow in 1993, which 1990s, this trend is not expected to continue was the 5th most intensive inflow during the past throughout the effective time-span of climate 130 years (Leppäranta and Myrberg 2009). change. There are not many candidates for the driving force behind the increasing trend in the DEEP 246 Raateoja • Boreal Env. Res. vol. 18

100 the BS is not a novel observation; to upscale, ) 3 80 this has been going on more or less constantly

60 for the past century (Fonselius and Valderrama 2003), and an adverse change of at least a similar 40 magnitude to the one presented here occurred 20

Riverine inflow (km between 1910 and 1930. The interpretation of a long environmental dataset frequently ignores 0 1970 1975 1980 1985 1990 1995 2000 2005 2010 the fluctuations found on a time-scale shorter Fig. 6. River inflow to the Gulf of Bothnia. than the actual span of the data. In this article, I concentrate on recently-occurring fluctuations over the period of a decade or two. salinity since the low-point in the early 1990s. Hypoxia does not seem to pose any immedi- As the river inflow does not show any apparent ate threat to the offshore BS environment. When trend at this time (Fig. 6), the driver has to be the available data since the low-point in the either the increased salinity of the deep inflow DEEP salinity in the early 1990s are extrapolated entering the BS or the increased volume of the to the future, hypoxic conditions — frequently –1 –1 deep water inflow. In this case, the former seems defined as < 2 mg O2 l or < 1.4 ml O2 l to be the more plausible option; the increasing (Conley et al. 2009) — would not be observed DEEP density in the NBP and ÅS shown in Fig. near the bottom before the 2060s, regardless 5B was a manifestation of the increasing salinity of the type of approach. By that time, climate (data not shown). change will affect the BS system, e.g., the water balance of the BS, in ways that are difficult to predict at this stage (BACC 2008). Thus, the pre-

Past and future of the O2 conditions in dictions presented here have to be regarded only the BS as a first-order estimate based on current knowledge. However, they still suggest that profound The linkage between the BS and the NBP is changes to the hydrographic regime of the BS obscured by the topographic and hydrographic and/or to the anthropogenic load on the BS will features of the ÅS. Here, I describe what the have to take place if any hypoxic events are to existence of the topographic and hydrographic occur in the offshore BS in the near future. features in the ÅS area actually means to the BS If, however, hypoxic events do in fact emerge in terms of the O2 dynamics. in the BS at some level, the hypoxic conditions at This study relied on the data collected by the sediment–water interface will have adverse conventional monitoring programs. Thus, it is impacts on the benthic fauna that is currently right to ask whether this approach produced thriving in the BS (HELCOM 2009a). Hypoxic trends that could describe the true natural fluc- events would not have any effect on the phospho- tuations in adequate detail for drawing reliable rus cycle of the BS per se. For this to happen, the conclusions regarding the state of the BS. In my near-bottom O2 conditions should reach anoxic opinion, the dataset was of sufficiently high- levels. The pivotal geochemical consequence of quality and dense to describe the O2 regime of anoxia is liberation of the bio-available phos- a sea area that is less dynamic with respect to phorus in the benthic system, i.e., the internal

O2 than, e.g., the GOF. Particularly, the absence phosphorus loading (Conley et al. 2009). In the of direct hydrodynamic forcing of the NBP, that GOF, this process has led to a situation where take place in the GOF, is the key for the more the sediment processes largely control the water- stable and predictable O2 regime in the BS. column phosphorus stock, and where phosphorus efflux from the sediments has been suggested to have triggered the extensive cyanobacterial

Deep-water O2 conditions blooms in the late 1990s and early 2000s despite marked decreases in the external phosphorus

The worsening of the DEEP O2 conditions in loading to the GOF in the 1990s (Pitkänen et Boreal Env. Res. vol. 18 • Deep-water oxygen conditions in the Bothnian Sea 247

6 al. 2001, Pitkänen et al. 2003). The accelerated ) LL7 –1 internal phosphorus loading in the GOF was l SR5 a consequence of both the high allochthonous nutrient load leading to a high autochthonous 4 load of organic matter to the benthic system, as well as the GOF’s tendency to display strong vertical density gradients. The BS has got neither 2 -P concentration (µmol of these. The share of organic matter in the BS’s 4 sediments is less than half of that in the GOF’s PO et al. sediments (Lehtoranta 2008). The halocline 0 in the BS is sufficiently strong to restrict vertical 1990 1995 2000 2005 2010 2015 mixing only in the central basin, and even there Fig. 7. The temporal variation in the near-bottom PO4-P periods of mixing right down to the sea floor may concentration at LL7 in the GOF and SR5 in the BS. occasionally take place (Kullenberg 1981, Lep- Only data collected from the discrete samples taken ≤ 5 m above the seafloor were included. päranta and Myrberg 2009). As a result, internal

P loading has not occurred in the BS, and PO4-P concentrations in the near-bottom layers of the

BS have not increased recently, which sets the BS back to this O2 supply is that it is meditated by apart from the GOF (Fig. 7). To conclude, there the flow of water of higher salinity than the BS is no reason to expect that a similar kind of cause- water, which strengthens the density gradient in and-effect cascade of eutrophication — increased the deep layer and restricts the vertical O2 flow. autochthonous load of organic matter — anoxia When the near-bottom density data are post- — internal loading of phosphorus that has pro- poned in time by two years, the near-bottom gressed in the GOF could take place in the BS. density pattern closely resembles a mirror image

Considering the utmost importance of this phos- of the near-bottom O2 saturation pattern (Fig. phorus source to the past adverse development of 4A). The hypotheses to explain the observed the GOF, it is apparent that the offshore BS will relation are as follows: (i) the organic matter — not follow the path of the GOF in this respect. descending to the sea floor and gradually being

decomposed by O2-consuming microbial proc-

esses — is a constant sink of O2. Both a lateral

Vertical and lateral O2 flows shaping the and a vertical supply of O2 are required to com- deep O2 conditions pensate this consumption in order to retain the

good O2 conditions that are typically met with in In the Baltic Sea, water density variation is the deep layer of the BS. (ii) Of the supplies of largely dictated by salinity variation (Kullen- O2, the lateral one can be considered a constant berg 1981); this fact has also been verified by source. Thus, it retains the “baseline” O2 condi- this study. In the Baltic Sea, the salinity-driven tions in the deep layers. The vertical one, in vertical density gradient is strong and relatively turn, gets to some extent diminished whenever independent of the time of year, and generally the water-column density gradient gets stronger. serves as a stable and effective barrier to the Whenever the vertical O2 supply decreases, the vertical flow of O2. In the offshore BS, however, O2 conditions worsen, only not to the brink the salinity-driven vertical density gradient of of hypoxia. This could explain the temporally 0.5 to 1.5 kg m–3 is much weaker than in the NBP opposite behavior of the near-bottom density and

(Table 1), which leaves room for a vertical O2 O2 saturation patterns. (iii) The O2 conditions flow to be a more important O2 supply for the do not worsen abruptly after the decrease of the deep layer in the BS than in the more southerly vertical O2 supply. The remaining lateral O2 flow basins of the Baltic Sea (Table 1). The lateral compensates to some extent for the O2 demand deep water flow, originating from the ÅS and of the biological decomposition process, which ending up to form the deep water mass of the leads to the observed temporal lag between the

BS, also carries O2 to the deep waters. The draw- near-bottom density and O2 saturation patterns. 248 Raateoja • Boreal Env. Res. vol. 18

The coastal area future. For consistent hypoxic events to emerge, the climate of the northern hemisphere would The coastal area of the GOF is often a mosaic have to force profound changes in the hydro- of water and land, and in terms of water circu- graphic regime of the BS, and some of these lation, the coastal basins can be quite isolated changes are not predicted to happen in the time from the general offshore circulation pattern. In span of the on-going climate change. Neverthe- these coastal basins, O2 problems have occurred less, the decreasing O2 trends in the offshore for a long time, and their extent has lately Bothnian Sea warrant paying of close attention to increased, most probably due to eutrophication the future development of the BS environment. (Wallius 2006). Some scattered areas of the As the emergence of hypoxic events does not Finnish coastal zone of the BS also suffer from a seem likely in the near future, there is no reason restricted water exchange, and sporadic hypoxic to expect that the internal loading of phosphorus events have also been observed there (Lundberg will occur in the offshore Bothnian Sea. Consid- et al. 2009). However, this zone is rather narrow ering the utmost importance of this phosphorus and the area subject to hypoxia is rather lim- source to the past adverse development of the ited; even there, hypoxic events only have been Gulf of Finland, it is apparent that the Bothnian observed in the deepest of the locally-restricted Sea will not follow the example of the Gulf of basins (Conley et al. 2011). If assessed by the Finland with regard to eutrophication. level of degradation of the benthic system, the width of the coastal area in the BS chronically Acknowledgements: I am grateful to the anonymous review- influenced by river runoff and/or direct anthro- ers who steered me in the right direction with this article. I pogenic load extends only to about 10 km at also thank the SYKE Marine Research Centre and the erst- while Finnish Institute of Marine Research for the collection most, even off the largest towns or industrial of the marine monitoring data and the permission to use it. A. facilities (Sarvala and Sarvala 2005; my inter- Räike kindly provided the riverine monitoring data, and R. pretation based on the data). In the outer and King checked the language. exposed archipelago areas, wind-driven mixing and the cyclonic water circulation pattern of the BS (Myrberg and Andrejev 2006), and particu- References larly vertical mixing that is not suppressed by any salinity-driven density gradient (Table 1) Alheit J., Möllman C., Dutz J., Kornilovs G. & Loewe P. ensure that the water exchange remains at a level 2005. Synchronous ecological regime shifts in the central Baltic and in the North Sea in the late 1980s. ICES J. high enough to prevent the emergence of even Mar. Sci. 62: 1205–1215. episodic hypoxic events. Consequently, since the BACC 2008. Assessment of climate change for the Baltic

1990s, the observed winter O2 concentration in Sea basin. Regional climate studies. Springer, Berlin the DEEP at the coastal stations studied has con- Heidelberg. sistently fallen within the range of 7 to 9 ml l–1 Conley D.J., Björck S., Bonsdorff E., Carstensen J., Destouni G., Gustafsson B.G., Hietanen S., Kortekaas M., Kuosa (data not shown), as compared with the 4 to 5 –1 H., Meier H.E.M., Müller-Karulis B., Nordberg K., ml l of the DEEP at the offshore stations. To Norkko A., Nürnberg G., Pitkänen H., Rabalais N.N., conclude, in the Finnish coastal zone of the BS, Rosenberg R., Savchuk O.P., Slomp C.P., Voss M., Wulff F. & Zillen L. 2009. Hypoxia-related processes in the any O2 problems will most probably continue to be concentrated in the inner coastal areas. Baltic Sea. Env. Sci. Tech. 43: 3412–3420. Conley D.J., Carstensen J., Aigars J., Axe P., Bonsdorff E., Eremina T., Haahti B.-M., Humborg C., Jonsson P., Kotta J., Lännegren C., Larsson U., Maximov A., Conclusions Rodriquez Medina M., Lysiak-Pastuszak E., Remeikaitė- Nikienė N., Walve J., Wilhelms S. & Zillén L. 2011. Even though the oxygen conditions in the deep Hypoxia in increasing in the coastal zone in the Baltic layers of the offshore Bothnian Sea have some- Sea. Env. Sci. Tech. 45: 6777–6783. Feistel R., Nausch G. & Wasmund N. (eds.) 2008. State what deteriorated during the most recent two and evolution of the Baltic Sea, 1952–2005: a detailed decades, there is little evidence to suggest that 50-year survey of meteorology and climate, physics, hypoxic events could emerge there in the near chemistry, biology, and marine environment. John Wiley Boreal Env. Res. vol. 18 • Deep-water oxygen conditions in the Bothnian Sea 249

& Sons, Hoboken, New Jersey. Leppäranta M. & Myrberg K. 2009. Physical oceanography Fleming-Lehtinen V., Laamanen M., Kuosa H., Haahti H. & of the Baltic Sea. Springer-Praxis Books in Geophysical Olsonen R. 2008. Long-term development of inorganic Sciences, Praxis Publishing, Chichester, UK. nutrients and chlorophyll a in the open Northern Baltic Lundberg C., Jakobsson B.-M. & Bonsdorff E. 2009. The Sea. Ambio 37: 86–92. spreading of eutrophication in the eastern coast of the Fonselius S. & Valderrama J. 2003. One hundred years of Gulf of Bothnia, northern Baltic Sea — an analysis in hydrographic measurements in the Baltic Sea. J. Sea time and space. Est. Coast. Shelf Sci. 82: 152–160. Res. 49: 229–241. Marmefelt E. & Omstedt A. 1993. Deep water properties in Grasshoff K., Kremling K. & Ehrhardt M. (eds.) 1999. Meth- the Gulf of Bothnia. Cont. Shelf Res. 13: 169–187. ods of seawater analysis. Wiley-VCH, Weinheim. Millero F.J. & Poisson A. 1981. International one-atmos- HELCOM 2009a. Biodiversity in the Baltic Sea: An inte- phere equation of state of seawater. Deep-Sea Res. 28: grated thematic assessment on biodiversity and nature 625–629. conservation in the Baltic Sea. Balt. Sea Env. Prog. Myrberg K. & Andrejev O. 2006. Modelling of circulation, 116B. Baltic Marine Environment Protection Commis- water exchange and water age properties of the Gulf of sion. Bothnia. Oceanologia 48: 55–74. HELCOM 2009b. Eutrophication in the Baltic Sea — an Pitkänen H., Lehtoranta J. & Räike A. 2001. Internal nutrient integrated assessment of the effects of the nutrient fluxes counteract decreases in external load: The case of enrichment and eutrophication in the Baltic Sea region. the estuarial eastern Gulf of Finland, Baltic Sea. Ambio Balt. Sea Env. Prog. 115 B, Baltic Marine Environment 30: 195–201. Protection Commission. Pitkänen H., Lehtoranta J., Peltonen H., Laine A., Kotta J., Hietala R., Lundberg P. & Nilsson J.A.U. 2007. A note on Kotta I., Moskalenko P., Mäkinen A., Kangas P., Perttilä the deep-water inflow to the Bothnian Sea. J. Mar. Syst. M. & Kiirikki M. 2003. Benthic release of phospho- 68: 255–264. rus and its relation to environmental conditions in the Håkansson B., Alenius P. & Brydstren L. 1996. Physical estuarial Gulf of Finland, Baltis Sea, in the early 2000s. environment in the Gulf of Bothnia. Ambio special Proc. Est. Acad. Sci. Biol. Ecol. 52: 173–192. report 8: 5–12. Sarvala M. & Sarvala J. (eds.) 2005. Miten voit Selkämeri? ICES 2008. Report of the working group on integrated Report series Ympäristön tila Lounais-Suomessa, 4. assessment of the Baltic Sea (WGIAB). ICES VGIAB South-West Finland Environmental Centre. report 2008, International Council for the Exploration Selman M., Greenhalgh S., Diaz R. & Sugg Z. 2008. of the Sea. Eutrophication and hypoxia in coastal areas: a global Jaanus A., Andersson A., Olenina I., Toming K. & Kaljurand assessment of the state of knowledge. WRI policy note, K. 2011. Changes in phytoplankton communities along a World Resources Institute. north-south gradient in the Baltic Sea between 1990 and Stigebrandt A. & Gustafsson B.G. 2003. Response of the 2008. Boreal Env. Res. 16A: 191–208. Baltic Sea to climate change — theory and observations. Kraberg A.C., Wasmund N., Vanaverbeke J., Schiedek D., J. Sea Res. 49: 243–256. Wiltshire K.H. & Mieskowska N. 2011. Regime shifts in UNESCO 1980. Tenth report of the joint panel on oceano- the marine environment: the scientific basis and political graphic tables and standards. UNESCO Tech. Pap. Mar. context. Mar. Poll. Bull. 62: 7–20. Sci. 36, UNESCO, Sidney, B.C., Canada. Kullenberg G. 1981. Physical oceanography. In: Voipio, A. Wallius H. 2006. Permanent seafloor anoxia in coastal basins (ed.), The Baltic Sea, Amsterdam Elsevier Oceanogra- of the northwestern Gulf of Finland, Baltic Sea. Ambio phy Series 30, Elsevier, Amsterdam, pp. 135–181. 35: 105–108. Lehmann A., Getzlaff K. & Harlaß J. 2011. Detailed assess- Winsor P., Rodhe J. & Omstedt A. 2001. Baltic Sea ocean ment of climate variability in the Baltic Sea for the climate: an analysis of 100 yr of hydrographic data with period 1958 to 2009. Clim. Res. 46: 185–196. focus on the freshwater budget. Clim. Res. 18: 5–15. Lehtoranta J., Ekholm P. & Pitkänen H. 2008. Eutrophica- Zillen L., Conley D.J., Andrén T., Andrén E. & Björck S. tion-driven sediment microbial processes can explain the 2008. Past occurrences of hypoxia in the Baltic Sea and regional variation in phosphorus concentration between the role of climate variability, environmental change and Baltic Sea sub-basins. J. Mar. Syst. 78: 495–504. human impact. Earth Sci. Rev. 91: 77–92.