Journal of Marine Systems 141 (2015) 190–199

Contents lists available at ScienceDirect

Journal of Marine Systems

j ourna l homepage: www.e lsev ie r .com/ locate / jmarsys

Biogeochemical and environmental drivers of coastal hypoxia

Angela M. Caballero-Alfonso a,1, Jacob Carstensen b, Daniel J. Conley a

a Department of Geology, Lund University, Sölvegatan 12, SE-223 62 Lund, Swedenb Department of Bioscience, Aarhus University, Frederiksborgvej 399, DK-4000 Roskilde, Denmark

E-mail address: angela.caballero-alfonso@geol.lu.se (A1 Tel.: +46 462220449.

http://dx.doi.org/10.1016/j.jmarsys.2014.04.0080924-7963/© 2014 The Authors. Published by Elsevier B.V

a b s t r a c t

a r t i c l e i n f o

Article history:Received 9 January 2014Received in revised form 31 March 2014Accepted 11 April 2014Available online 21 April 2014

Keywords:HypoxiaCoastal zoneBaltic SeaNutrients

Recent reports have demonstrated that hypoxia is widespread in the coastal zone of the Baltic Sea. Here weevaluate the long-term trends of dissolved oxygen in bottomwaters and of the drivers of coastal hypoxia. Elevenof the 33 sites evaluated had increasing trends of bottom water dissolved oxygen, but only the StockholmArchipelago presents a consistent positive increasing trend in time. The vast majority of sites continue toworsen, especially along theDanish and Finnish coasts, in spite of remediation efforts to reduce nutrients. Surfacetemperatures were relatively comparable across the entire coastal Baltic Sea, whereas bottom water tempera-tures varied more strongly among sites, most likely due to differences in mixing (or stratification) and waterexchange with the open Baltic Sea. Nutrient concentrations varied by factors 2–3 with highest levels at siteswith restricted water exchange and higher land based nutrient loading. None of the sites were permanentlystratified during the summer seasonal window although most of the sites were stratified more than half of thetime. The frequency of hypoxia was also quite variable with sites in Gulf of Bothnia almost never experiencinghypoxia to enclosed sites with more than 50% chance of hypoxia. There are many factors governing hypoxiaand the complexity of interacting processes in the coastal zone makes it difficult to identify specific causes.Our results demonstrate that managing nutrients can create positive feedbacks for oxygen recovery to occur.In the absence of nutrient reductions, the recovery from hypoxia in coastal marine ecosystems is unlikely.

© 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Hypoxia has intermittently occurred in the Baltic Sea since its forma-tion 8000 years ago with its occurrence constrained by hydrologicalprocesses, climate and anthropogenic activities (Sohlenius et al., 2001;Zillén et al., 2008). Since the 1950s the spatial distribution and intensityof hypoxia in the open waters have increased due to anthropogenicnutrient enrichment (Carstensen et al., 2014; Conley et al., 2009a,2009b). Nutrient loads have increased by approximately 2.5 times fornitrogen (N) and 3.7 for phosphorus (P) since the turn of the last century(Gustafsson et al., 2012). Nutrients stimulate phytoplankton productionand the subsequent decomposition of organic matter creates a demandfor oxygen. The Baltic Sea is comprised of several connected basins withpermanent salinity stratification and restricted ventilation of the bottomwatersmaking it susceptible to hypoxia. Hypoxia reduces the abundanceof benthic fauna and alters sediment nutrient dynamics (Conley et al.,2007; Karlson et al., 2007; Perus and Bonsdorff, 2004). Hypoxia also re-stricts the reproduction of the bottom dwelling fish species, especiallycod, by reducing spawning area (Limburg et al., 2011; Vallin et al., 1999).

Much less is known regarding hypoxia in the coastal zone, where thephysical processes are generally more dynamic and complex. Globally, the

.M. Caballero-Alfonso).

. This is an open access article under

ratesofdecrease inbottomwateroxygenconcentrationsarehigher in coast-al areas than in the open oceans (Gilbert et al., 2010). Hypoxia has been ob-served in the coastal zone of the Baltic Sea with 115 sites out of 613 sites,assessed from monitoring data, having experienced hypoxia between1955 and 2009 (Conley et al., 2011). In addition to the existing coastal ma-rine sites worldwide that have reported hypoxia (Diaz and Rosenberg,2008), these 115 sites make the Baltic Sea the largest region in the worldthat suffers from hypoxia with over the 20% of the known areas worldwide(Conley et al., 2011). In addition, there is an increasing trend in the severityof hypoxia since the 1950s in many coastal sites of the Baltic Sea.

There are many factors governing hypoxia and the complexity ofinteracting processes in the coastal zone makes it difficult to identifyspecific causes. Anthropogenic drivers are primarily nutrient inputs,one of the common features of coastal eutrophication. Eutrophicationis defined as the organic enrichment of a system (Nixon, 1995), whichhas serious consequences for the marine environment including de-clines in water transparency, increasing algal blooms, loss of benthicvegetation and hypoxia among others (Nixon, 2009). There are alsonatural drivers of hypoxia: 1) hydrology (reduced horizontal advectionenhancing bottom-water residence time, stratification reducing verticalmixing), and 2) climate (temperature reducing the solubility of oxygenand increasing respiration as well as changes in wind patterns, rainfalland runoff affecting the hydrological processes). In the open ocean cli-mate models predict declines in dissolved oxygen produced by global

the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

191A.M. Caballero-Alfonso et al. / Journal of Marine Systems 141 (2015) 190–199

warming due to a contribution of thermal, dynamical, and biogeochem-ical factors (Stramma et al., 2008). However, further studies are neededto understand and elucidate the underlying mechanisms for hypoxia incoastal areas (Conley et al., 2009b; Rabalais et al., 2010).

We focused our investigation on evaluating the long-term trends ofbottom water oxygen concentrations and of the drivers of coastal hyp-oxia in the Baltic Sea. Our goal was to examine relationships betweennutrients, physical factors and climate as important drivers for coastalhypoxia in the Baltic Sea.

2. Material and methods

2.1. Study area and site descriptions

The Baltic Sea is a nearly landlocked sea located between 54 and66°N having restricted water exchange with the oceans. It consists ofthe main deeper basin (Baltic Proper) with three major gulfs in thenortheast (Gulf of Riga, Gulf of Finland and Gulf of Bothnia) and con-nects to the North Sea through the Danish Straits. Most of the BalticSea is brackish with salinities less than 10, but the entrance area (theDanish Straits) is characterized by strong salinity gradients. The climateis temperate with an annual rainfall of approximately 600 mm per year(Omstedt et al., 1997). About 85 million people inhabit the Baltic Seacatchment, which is dominated by agriculture and urbanization in thesouthwest and forests towards the northeast. Nutrient loads from thedifferent regions are a reflection of the land use (Wulff et al., 1990).

The coastal zone surrounding the Baltic Sea is quite diverse, owing toits geologic history. Shallow, sandy estuaries and coastal embaymentswith complex geomorphology dominate in the southwest, changing toan open coastline with lagoons along the southeastern coastline. Mostof the Swedish and Finnish coastline is archipelagic. Large rivers drainan area approximately four times the Baltic Sea itself and discharge tothe Baltic Proper, Gulf of Riga, Gulf of Finland and Gulf of Bothnia. Riversdraining into the Danish Straits are smaller.

A total of 33 sites were studied and grouped in 9 regions followingHELCOM designations: Kattegat (4 sites), Øresund (1 site), Belt Sea(6 sites), West Gotland Basin (4 sites), Stockholm Archipelago (5 sites),Bothnian Sea (1 site), Bothnian Bay (2 sites), Finnish Archipelago (5sites) and Gulf of Finland (5 sites); see Table 1 for more details. Thesesites were selected based on the results in Conley et al. (2011) and thelength of time series available in the Baltic Environmental Database(BED). We chose sites that had experienced coastal hypoxia and wererepresentative of a region, and which had more than 10 years of data.Some data-rich areas, e.g. the Stockholm Archipelago, had numeroussites with long-term time series and in such areas the most representa-tive sites were selected.

2.2. Data analyses

Seasonal windows for hypoxia-prone months were selected perregion to give a better description of low oxygen conditions (Conleyet al., 2011). For the Bothnian Bay, Stockholm and Finnish archipelagos,the West Gotland Basin, the Kattegat and the Öresund the summermonths considered were August through October. For the BothnianSea we considered only August and September; and for the Gulf ofFinland July through October. These data were organized by regionand sites with the aim to establish annual time series for each site.

In coastal areas periods of stratification and mixing can alternateover short time scales, unlike the hydrographic conditions in the deeperBaltic Sea basins (Conley et al., 2011). Stratification is a precondition forhypoxia to occur and therefore only profiles characterized by stratifiedconditions were selected to evaluate oxygen trends. We defined thewater column to be stratified if the density difference between surfaceand bottom waters was larger than 0.5 g l−1 (see Conley et al., 2007).The rate of oxygen depletion in the bottom water depends on the vol-ume below the pycnocline and to characterize this volume we

calculated bottom water thickness as the profile maximum depthminus the depth location of the pycnocline. This characteristic was cal-culated on the more recent data, due to low resolution of older profiles,and assuming that bottomwater thickness remained the same through-out the summer period.

Oxygen saturation (%) and Apparent Oxygen Utilization (AOU)werecalculated for the bottom waters. AOU is defined as the observed oxy-gen concentration subtracted from the oxygen concentration at satura-tion state calculated from temperature and salinity of the bottomwater(Hetland and Di Marco, 2008). AOU allows temperature and salinityeffects on oxygen solubility to be removed from the observations andquantifies the loss of oxygen due to biological and chemical activities(Bianchi et al., 2010), most importantly respiratory demands fromdecomposition of organicmaterial. The bottomoxygen trendwas testedagainst the following possible hypoxia triggers: (a) physical status:temperature and stratification; (b) biological status: surface chlorophylla (chla); and (c) chemical status: total phosphorus (TP) and total nitro-gen (TN). Additionally, to assess the putative effect of low oxygen con-centrations on benthic communities the frequency of hypoxia wasassessed. Although the thresholds for faunal effects to occur varywidelyin the literature from 0.28 to 4 mg O2 l−1 (Steckbauer et al., 2011;Vaquer-Sunyer and Duarte, 2010), we define hypoxia as concentrationsof dissolved oxygen equal to or below 2 mg O2 l−1 (Rabalais et al.,2010). Hence, the frequency of hypoxia was calculated as the propor-tion of profiles with bottom water oxygen concentrations below 2 mgO2 l−1.

Trends of seasonal means were estimated using Generalized LinearModels (GLMs), separately for each site and variables of interest, takinginto account temporal variations (months and years as additive categor-ical factors). After examining the distribution of different variables,oxygen concentrations, AOU, and temperature were assumed to followa normal distribution; chla, TN and TP were assumed lognormal distrib-uted; and frequencies of hypoxia and stratificationwere binomial. Usingthe GLM approach accounted for changes in sampling across years byweighting each profile according to the month it was taken (Carstensenet al., 2006). Trends in oxygen concentrations were related to trends intemperature, frequency of stratification, surface TN, TP and chla concen-trations using Spearman correlations. Spearman correlation was chosensince it is less sensitive to single years with strongly deviating annualmeans that may otherwise influence standard correlations. All the dataprocessing and statistical analyses were programmed using SAS®.

3. Results

The studied sites varied from 10 to 117 m depth with bottom waterbelow the pycnocline ranging from 1.1 to 22.2 m (Table 1). The thick-ness of the bottom water typically increased with water depth. Surfacetemperatureswere relatively comparable across the entire coastal BalticSea, whereas bottom water temperatures varied more strongly amongsites, most likely due to differences in mixing (or stratification) andwater exchange with the open Baltic Sea. Nutrient concentrations var-ied by factors 2–3 with the highest levels at sites with restricted waterexchange and higher connectivity to land. None of the sites were per-manently stratified during the summer seasonal window althoughmost of the siteswere stratifiedmore than half of the time. The frequen-cy of hypoxiawas also quite variablewith sites in Gulf of Bothnia almostnever experiencing hypoxia to enclosed sites with more than 50%chance of hypoxia.

In the following trends in oxygen conditions and associated vari-ables are presented, but the trends are only visualized for a few selectedsites for plotting, aswe could not showall. However, directions of trendsare visualized geographically for all sites (Fig. 1). The sites presented(Figs. 2–5) were those having the highest significant correlation forbottom oxygen concentration versus at least one of the tested influenc-ing variables (Table 2).

Table 1Characteristics of the 33 coastal study sites in the Baltic Sea. Site names are from the nationalmonitoring programs. Number of profiles anddescriptive statistics characterizing thesewere calculated for the seasonalwindowas described in theMaterialand methods section. (Abbreviations: Max. depth—maximum depth (m); bottom layer thickness (m); temperature (°C); TN— total nitrogen (μM); TP— total phosphorus (μM); Surf— surface water samples; Bot— bottomwater samples; AOU—

Apparent Oxygen Utilization; Freq. — frequency).

Region Site Max. depth # of profiles # of bottomoxygen samples

Bottom layerthikness

Temperature TN TP Bottom AOU Freq. ofstratification (%)

Freq. ofhypoxia (%)

Surf Bot Surf Bot Surf Bot

Gulf of Bothnia U-5 Uudenkaarlepyyn 23 551 550 10.4 11.6 10.2 33.4 31.4 0.7 0.6 1.1 16.5 0.2U-6 Uudenkaarlepyyn 15 103 99 7.6 14 12.3 40.1 27.8 0.9 0.6 1.1 40.8 4.9Pome 115 Preiviikinlahti 26 277 274 9.8 14.7 11 20.6 18.1 0.4 0.4 2.3 36.8 4.0

Finnish Archipelago Hala N 22 114 114 1.3 14.4 9.3 38 50.8 1.1 1.6 – 32.5 48.3Kamsholmsfjärden 19 178 176 2.5 17.7 8.4 27.8 50.9 0.9 7.6 10.5 89.9 80.3Pala 47 223 223 18.5 17.6 12.7 30.2 31.7 1 1.3 4.5 52.9 3.1Turm 290 Kuparivuori 25 83 82 1.2 15.7 9.5 30.5 39.7 1 1.4 5.6 54.2 22.9Uki 245 Vähä-Seikomaa 26 1007 1002 10.5 13.7 12.5 28.7 29 1.2 1.6 3 17.8 4.2

Gulf of Finland Orrenkylänselkä 34 356 354 10.1 16.3 10.6 40.2 35.4 1.9 3 4.9 61.0 4.8Skatafjärden 28 90 90 3.2 16.4 8.9 25.6 27.9 0.9 2 4.5 88.9 2.2Suomenl Varissaari 17 193 193 2.1 15.9 10.9 34.5 30.9 1.1 1.6 3.5 65.8 2.6Svartbäckinselkä 42 267 265 9.7 15.7 7.4 29.3 29.4 1.2 1.9 4.7 75.3 1.5UUS-18 Sandöfjärden 31 301 301 8 15.9 8.7 26.7 34.5 1.1 4.4 7.2 71.1 38.5

Stockholm Archipelago Blomskär 27 153 136 2.9 13.8 5.2 49.5 52.8 2 5.4 11 90.2 72.6Kanholmsfjärden 117 164 135 17.4 13.7 5.4 20.4 34.6 0.6 4.8 10.3 79.9 59.2Karantänbojen 22 157 153 4.1 13.4 6.1 56.2 57.4 2.2 4.6 9.6 91.7 59.9Solöfjärden 47 357 312 14 13.4 7.5 39.4 32.3 1.4 2.4 6.5 84.3 16.8Vaxholmsfjärden 29 240 226 5.3 14 8.9 40.6 50 1.5 3.9 8.3 72.1 61.3

West of Gotland Basin Blankaholm 24 28 28 2.4 15.3 5.9 29.8 33.7 0.9 5.4 9.4 75.0 46.4Bråvik 33 70 69 11.1 16.7 11.1 36 29.9 1.1 1.9 5.1 77.1 4.3Västervik 64 90 89 22.2 15.9 4.8 30.5 50.6 0.9 5.4 9.3 83.3 45.6Västrum 17 28 28 1.4 15.3 7.5 32 45.6 1.1 9 9.4 78.6 71.4

Belt Sea Aabenraa Fjord 35 235 235 5.1 16.2 11 27.8 34.8 1.4 4.5 7.3 98.3 69.4Det Sydfynske Øhav 40 821 808 15 15.5 13.6 22.1 27.4 0.9 2.3 3.5 64.2 28.6Eckernförder Bucht 31 80 72 5.2 15.5 11.6 21.3 26.9 1.2 4.1 7.8 82.5 73.8Flensborg Fjord 31 345 343 7.7 16.2 11.1 27.4 40.6 1.6 5.4 7.6 91.6 69.6Horsens Fjord 23 497 489 11.8 15.2 14.5 28.4 28.3 1.8 2 2.4 63.2 6.4Århus Bugt 18 333 333 1.7 15.1 13 15.9 17.9 0.8 1.5 4.6 95.8 13.2

Kattegat and Öresund Hevring Bugt 10 151 151 1.1 14.6 14 16.4 17.5 0.8 1.2 2.2 78.1 2.7Laholmsbugten 28 127 123 11.3 16.4 13 20.5 20.3 0.5 0.9 3.7 82.7 16.5Middelgrund 19 198 198 2.2 14.9 12.5 19.5 21 0.6 1.1 4.2 95.0 7.1Skälderviken 29 98 93 14.2 16.5 13.9 25 21.9 0.6 1.1 2.8 79.6 10.2Valö 24 74 72 2.6 16.1 13.5 18.1 19.7 0.6 1.1 3.1 90.5 4.1

192A.M

.Caballero-Alfonso

etal./JournalofMarine

Systems141

(2015)190

–199

Fig. 1.Maps summarizing the trends for the bottomoxygen concentrations, AOU, temperature, frequency of stratification, TN and TP. The regions are grouped as: 1) Gulf of Bothnia,2) Finnish Archipelago and 3)Gulf of Finland, 4) StockholmArchipelago, 5)West GotlandBasin, 6) Belt Sea and 7)Kattegat andÖresund. Increasing and decreasing trends are shownwithupward and downward facing triangles, green or red colors indicating improving or worsening conditions, and open and filled triangles signify non-significant and significant trends,respectively.

193A.M. Caballero-Alfonso et al. / Journal of Marine Systems 141 (2015) 190–199

3.1. Gulf of Bothnia region

Three sites were investigated in this region (U-5 Uudenkaarlepyyn,U-6 Uudenkaarlepyyn and Pome 115 Preiviikinlahti); two of these aregeographically close but have different nutrient levels and frequencyof stratification (Table 1). Pome 115 Preiviikinlahti is a relatively deep

site (26 m) with temperature stratification of ~4 °C separating a rela-tively large volume of bottom water. Compared to the other sitesfrequencies of stratification and hypoxia were both relatively low. Thethree sites had no clear distinctive trends in oxygen conditions, temper-ature, nutrient levels or stratification patterns, but there were stronginterannual variations in temperature and frequency of stratification

0

1

2

3

4

5

6

70

2

4

6

8

10

12

AO

U (

mg/

l)

Bot

tom

Oxy

gen

(mg/

l)B

otto

m te

mpe

ratu

re (

°C)

U-5 UudenkaarlepyynBottom oxygenPome 115 PreiviikinlahtiBottom oxygenU-5 UudenkaarlepyynAOUPome 115 PreiviikinlahtiAOU

A

0

10

20

30

40

50

60

70

80

90

100

0

2

4

6

8

10

12

14

16

18

20

Fre

quen

cy o

f str

atifi

catio

n (%

)

U-5 UudenkaarlepyynBottom temperaturePome 115 PreiviikinlahtiBottom temperatureU-5 UudenkaarlepyynFreq. of stratificationPome 115 PreiviikinlahtiFreq. of stratification

B

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0

10

20

30

40

50

60

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

2008

2011

TP

(µM

)

TN

(µM

)

Year

U-5 Uudenkaarlepyyn TNPome 115 Preiviikinlahti TNU-5 Uudenkaarlepyyn TPPome 115 Preiviikinlahti TP

C

Fig. 2. Long-term trends in the Gulf of Bothnia exemplified by two sites. A: dissolvedbottom oxygen concentrations and Apparent Oxygen Utilization; B: bottom temperatureand frequency of stratifications; C: surface TN andTP. Yearly seasonalmeanswere obtainedfrom the GLM approach.

Bot

tom

Oxy

gen

(mg/

l)B

otto

m te

mpe

ratu

re (

°C)

TP

(µM

)

TN

(µM

)

0

2

4

6

8

10

120

1

2

3

4

5

6

7

8

9

AO

U (

mg/

l)

Kamsholmsfjärden Bottom oxygenOrrenkylänselkä Bottom oxygenKamsholmsfjärden AOUOrrenkylänselkä AOU

A

0

10

20

30

40

50

60

70

80

90

100

0

2

4

6

8

10

12

14

16

Fre

qu

ency

of

stra

tifi

cati

on

(%

)

Kamsholmsfjärden Bottom temperatureOrrenkylänselkä Bottom temperatureKamsholmsfjärden Frequency of stratificationOrrenkylänselkä Frequency of stratification

B

0

1

2

3

4

5

6

7

0

10

20

30

40

50

60

70

80

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

2008

2011

Year

Kamsholmsfjärden TNOrrenkynlänselkä TNKamsholmsfjärden TPOrrenkynlänselkä TP

C

Fig. 3. Long-term trends in the Finnish Archipelago and the Gulf of Finland exemplified bytwo sites. Details as for Fig. 2.

194 A.M. Caballero-Alfonso et al. / Journal of Marine Systems 141 (2015) 190–199

(Fig. 2). However, Pome 115 Preiviikinlahti experienced a period withlow oxygen concentrations in the 1980s, but these low concentrationscould not be linked to any of the potential explanatory variables(Table 2). Temperature was important in explaining trends in oxygenconditions at the two Uudenkaarlepyyn sites. For U-6 Uudenkaarlypyynchanging stratification pattern was also important (Table 2), whereasthis factor was not significant at U-5 Uudenkaarlepyyn that generallyhad a higher mixing frequency (Table 1).

3.2. Finnish Archipelago region

This region includes five sites (Hala N, Kamsholmsfjärden, Pala,Turm 290 Kuparivuori and Uki 245 Vähä-Seikomaa) located along theinner coastal archipelago with relatively high nutrient levels comparedto the other sites (Table 1). Overall, the frequency of stratification aver-aged 44%, but there were large differences in stratification betweensites. Temperature differences between surface and bottom were largeat all sites except Uki 245 Vähä-Seikomaa, and bottom water tempera-tures have been increasing for the last two decades (Fig. 3B). Frequencyof hypoxia was even more variable ranging from 3.1% to 80.3% with anaverage of 31.8%. Three sites had a small bottom water volume and

these three siteswere also the siteswith thehighest frequencyof hypoxia.There were no distinctive trends in the variables (most were not signif-icant, Fig. 1), although there were large interannual variations (Fig. 3).Significant correlationswere obtained between the bottomoxygen con-centration and the bottom temperature (Table 2).

3.3. Gulf of Finland region

Five sites were investigated in this region (Orrenkylänselkä,Skatafjärden, Suomenl Varissaari, Svartbäckinselkä and UUS-18Sandöfjärden) and all of them were stratified by temperature through-out most of the summer period (Table 1). Stratification patterns variedsubstantially between years. Temperature differences were large sepa-rating a cold bottom water mass (7–11 °C) of several meters from awarmer surface layer (~16 °C). The last two decades have generallybeen warmer. Nutrient levels were relatively high and did not changeover the years (Fig. 3C). Despite strong stratification the frequency ofhypoxia was low, except for UUS-18 Sandöfjärden (Table 1). This siteis a semi-enclosed embayment where bottom water ventilation couldbe low. Oxygen concentrations correlated significantly with tempera-ture and/or nutrient concentrations at the five sites, displayingworsen-ing oxygen conditions with increasing temperature and nutrients(Table 2).

Bot

tom

Oxy

gen

(mg/

l)B

otto

m te

mpe

ratu

re (

°C)

TN

(µM

) 0

2

4

6

8

10

120

1

2

3

4

5

6

7

8

9

AO

U (

mg/

l)

Vaxholmsfjärden Bottom oxygenVästervik Bottom oxygenVaxholmsfjärden AOUVästervik AOU

A

0

10

20

30

40

50

60

70

80

90

100

0

2

4

6

8

10

12

14

16

Fre

quen

cy o

f str

atifi

catio

n (%

)Vaxholmsfjärden Bottom temperatureVästervik Bottom temperatureVaxholmsfjärden Frequency of stratificationVästervik Frequency of stratification

B

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

0

10

20

30

40

50

60

70

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

2008

2011

TP

(µM

)

Year

Vaxholmsfjärden TN Västervik TNVaxholmsfjärden TP Västervik TP

C

Fig. 4. Long-term trends in the StockholmArchipelago andWest GotlandBasin exemplifiedby two sites. Details as for Fig. 2.

Bot

tom

Oxy

gen

(mg/

l)B

otto

m te

mpe

ratu

re (

°C)

TN

(µM

)

0

1

2

3

4

5

6

7

8

9

100

1

2

3

4

5

6

7

8

9

10

AO

U (

mg/

l)

Aabenraa Fjord Bottom oxygenHorsens Fjord Bottom oxygenAabenraa Fjord AOUHorsens Fjord AOU

A

0

10

20

30

40

50

60

70

80

90

100

0

2

4

6

8

10

12

14

16

18

Fre

quen

cy o

f str

atifi

catio

n (%

)

Aabenraa Fjord Bottom temperatureHorsens Fjord Bottom temperatureAabenraa Fjord Frequency of stratificationHorsens Fjord Frequency of stratification

B

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

0

10

20

30

40

50

60

70

80

90

100

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

2008

2011

TP

(µM

)

Year

Aabenraa Fjord TN

Horsens Fjord TN

Aabenraa Fjord TP

Horsens Fjord TP

C

Fig. 5. Long-term trends in the Danish strait exemplified by two sites. Details as for Fig. 2.

195A.M. Caballero-Alfonso et al. / Journal of Marine Systems 141 (2015) 190–199

3.4. Stockholm Archipelago region

In this region five sites were analyzed (Blomskär, Kanholmsfjärden,Karantänbojen, Sölofjärden and Vaxholmsfjärden). All sites wererelatively deep with Kanholmsfjärden as the deepest investigated ofall sites (Table 1). The bottom water volume below the pycnoclinewas relatively large; however, all sites were hypoxic for most of thesummer period. Nutrient levels in the Stockholm Archipelago werethe highest of all sites. Stratification was strong and driven by bothsalinity and temperature differences. Oxygen concentrations have in-creased and nutrient concentrations have decreased over time (Figs. 1,4A,C). The bottom water has generally warmed and the frequency ofstratification has decreased in Vaxholmsfjärden (Fig. 4B), but this pat-tern is not consistent at all sites. Bottom oxygen concentrations weresignificantly correlated with nutrients at Karantänbojen, Sölofjärdenand Vaxholmsfjärden, as expected, whereas oxygen conditionswere related to frequency of stratification in Kanholmsfjärden andpositively related to temperature in Vaxholmsfjärden. Whereas thepositive correlation to temperature in Vaxholmsfjärden could be anartifact from nutrients showing similar trends, the positive correla-tion to frequency of stratification in Kanholmsfjärden is somewhatmore surprising. Blomskär, which is an isolated branch in the

Stockholm Archipelago, had a significant decreasing trend of bottomoxygen concentration with time, as opposed to all the other sitesdisplaying increasing trend (Fig. 1). The oxygen trend at Blomskärwas not significantly correlated to any of the explanatory variables(Table 2).

3.5. West Gotland Basin region

Four sites were investigated (Blankaholm, Bråvik, Västervik andVästrum) with depths between 17 and 64 m; the two deepest sitesalso having a high bottom water thickness (Table 1). Nutrient levelswere moderate compared to the other sites, whereas the frequency ofstratification was high. The frequency of hypoxia was high at all sites,except Bråvikwhich is dominated by the largeMotala River. Freshwaterinputs to the other sites are relatively smaller. The monitoring data setfrom this region is relatively weak compared to other regions withfewer years of data. Consequently, it ismore difficult to assess the trendsand correlate oxygen conditions with other factors. Nevertheless, oxy-gen conditions have improved at Blankaholm and Västervik (Figs. 1,4C) and these trends correlated significantly with TP levels and temper-ature (Table 2).

Table 2Spearman correlations between trends of oxygen concentrations and physical, chemical and biological conditions believed to influence oxygen conditions. The significance of thecorrelations is indicated as *: p b 0.05; **: p b 0.01; ***: p b 0.001. Trends used for the correlations are shown for U-5 Uudenkaarlepyyn and Pome 115 Preiviikinlahti (Fig. 1),Kamsholmsfjärden and Orrenkylänselkä (Fig. 2), Vaxholmsfjärden and Västervik (Fig. 3), and Aabenraa Fjord and Horsens Fjord (Fig. 4). The symbol ‘–’ indicates that there were no ortoo few annual values to calculate and test the correlation.

Region Site Surface chla Bottom temperature Frequency of stratification Surface TN Surface TP

Gulf of Bothnia U-5 Uudenkaarlepyyn 0.5 −0.83*** −0.02 0.24 −0.17U-6 Uudenkaarlepyyn −0.5 −0.55*** −0.7*** 0.06 0.22Pome 115 Preiviikinlahti – −0.1 0.14 0.36 −0.09

Finnish Archipelago Hala N – −0.47*** 0.24 −0.01 −0.32Kamsholmsfjärden – −0.1 −0.19 −0.14 0.03Pala −0.38 −0.62*** 0.1 0.09 0.17Turm 290 Kuparivuori – −0.54*** 0.03 −0.25 −0.03Uki 245 Vähä-Seikomaa – −0.41** 0.22 0.12 0.09

Gulf of Finland Orrenkylänselkä −0.5 −0.07 0.00 −0.21 −0.48***Skatafjärden – −0.73*** 0.03 −0.38** −0.33*Suomenl Varissaari – −0.69*** −0.05 −0.16 −0.1Svartbäckinselkä – −0.29 0.15 −0.15 −0.34**UUS-18 Sandöfjärden – −0.05 0.37 −0.61*** −0.68***

Stockholm Archipelago Blomskär −0.4 −0.25 0.18 −0.11 0.07Kanholmsfjärden −0.2 0.00 0.34** 0.15 0.02Karantänbojen −0.5 −0.04 0.19 −0.44*** −0.65***Solöfjärden −0.4 −0.08 −0.02 −0.48*** −0.45***Vaxholmsfjärden 0.3 0.5*** −0.1 −0.36** −0.47***

West of Gotland Basin Blankaholm 0.04 0.24 −0.39 0.24 −0.61**Bråvik – −0.6** 0.13 −0.23 −0.21Västervik −0.09 −0.08 0.02 0.12 −0.31Västrum −0.25 0.42 0.06 0.07 −0.33

Belt Sea Aabenraa Fjord 0.11 0.31 −0.32 0.21 0.3Det Sydfynske Øhav −0.06 −0.17 −0.13 −0.24 −0.02Eckernförder Bucht – 0.07 −0.05 0.28 0.05Flensborg Fjord 0.11 0.18 −0.23 0.36 0.15Horsens Fjord −0.28 −0.23 −0.41** −0.36** −0.16Århus Bugt −0.22 −0.04 −0.1 −0.63*** −0.46*

Kattegat and Öresund Hevring Bugt 0.12 −0.06 −0.11 −0.45 −0.21Laholmsbugten −0.17 −0.16 −0.16 0.13 −0.01Middelgrund −0.62 0.14 0.07 −0.33 −0.36Skälderviken 0.05 0.03 0.24 0.11 0.05Valö 0.002 −0.26 −0.05 0.003 0.38

196 A.M. Caballero-Alfonso et al. / Journal of Marine Systems 141 (2015) 190–199

3.6. Belt Sea region

The Belt Sea region includes 6 sites (Aabenraa Fjord, Det SydfynskeØhav, Eckernförder Bucht, Flensborg Fjord, Horsens Fjord and ÅrhusBugt) with depths ranging from 18 to 40 m (Table 1). It is a region(Table 1) with a relatively high frequency of stratification (83%), dueto the location of a halocline around 15–25 m. Nutrient levels arerelatively low, compared to sites in the more brackish part of the BalticSea, but there are differences among sites in the Belt Sea region associ-ated with proximity to land-based nutrient sources. Three of the sites inthe Belt Sea region (Aabenraa Fjord, Eckernförder Bucht and FlensborgFjord) have high frequencies of hypoxia (~70%) and these sites areconnected to the Little Belt where bottom water ventilation is lowest.The three other sites display more episodic events of hypoxia. Despitedecreases in both nitrogen and phosphorus concentrations over thelast many decades (Figs. 1, 5C) oxygen conditions have not generallyimproved, although there is a tendency for improvement at some sitesin the most recent years (Fig. 5A). As a consequence, a general patternof declining nutrients associated with improving oxygen conditionswas not found, and only Horsens Fjord and Århus Bugt showed signifi-cant correlations (Table 2). The lack of significant correlations in thisregion could be due to the lack of variability in oxygen concentrationsover time making the identification of such relationships difficult.

3.7. Kattegat and Öresund region

Five sites are included in this region (Hevring Bugt, Laholmsbugten,Middelgrund, Skälderviken and Valö); the three deepest of theselocated on the Swedish coast. The sites are generally shallower thanthose in other regions and the thickness of the bottom layer dependson the permanent halocline in the Kattegat, which is typically found

around 15–20 m depth (Andersson and Rydberg, 1988). However,windy conditionsmix down the surface layer andmay tilt the halocline.The sites therefore have a high frequency of stratification (Table 1). Thesites in this region have the lowest nutrient levels of all Baltic Sea coastalsites, because they are more influenced by water exchanges with theNorth Sea where both TN and TP levels are lower than in the Baltic.The frequency of hypoxia is generally low and these events have anepisodic nature (Conley et al., 2011). In this region oxygen concentra-tions have displayed non-uniform trends (Fig. 1), but major changeshave not been observed despite decreasing nitrogen levels (Fig. 5). Dueto the episodic nature of low oxygen conditions and the lack of largechanges in oxygen conditions no significant correlations were foundbetween oxygen concentration and explanatory variables (Table 2).

3.8. Comparison of sites across regions

The 33 coastal sites were exposed to large differences in physicalforcing and behaved differently to nutrient enrichment. We examinedthe effect of stratification patterns on hypoxia by plotting AOU versusthe frequency of stratification and thickness of bottom layer (Fig. 6).Oxygen consumption was clearly related to frequency of stratificationwith hypoxia rarely occurring at sites that were stratified less than50% of the seasonal window (Table 1). There were regional differencesin AOU versus frequency of stratification. Sites in Kattegat, Öresund,Belt Sea and Gulf of Finland generally had lower AOU compared tosites in archipelagic regions (Finnish Archipelago, StockholmArchipelagoand West Gotland Basin) with the same frequency of stratification(Fig. 6A). The relationship between AOU and bottom water thicknesswas more complex, although at some sites it appears as if oxygenconsumption could be higher if the bottom water volume below thepycnocline was relatively low (Fig. 6B).

0

2

4

6

8

10

12

0 20 40 60 80 100

AO

U (

mg

/l)

Gulf of BothniaFinnish ArchipelagoGulf of FinlandStockholm ArchipelagoWest Gotland BasinBelt SeaKattegat & Öresund

A

0

2

4

6

8

10

12

0 5 10 15 20 25

AO

U (

mg

/l)

Bottom layer thickness (m)

B

Frequency of stratification (%)

Fig. 6. Relationships between the AOU and the frequency of stratification (A) and thick-ness of the bottom layer (B).

197A.M. Caballero-Alfonso et al. / Journal of Marine Systems 141 (2015) 190–199

4. Discussion

4.1. Overall trends

Reports of hypoxia in coastal marine systems have increased duringthe last few decades with the primary cause of low oxygen conditionslinked to increases in nutrient loads (Conley et al., 2011; Diaz andRosenberg, 2008; Rabalais et al., 2014). There are numerous studiesshowing that eutrophication is promoting the expansion, duration, andintensity of hypoxic events (Diaz and Rosenberg, 2008; Kemp et al.,2009; Rabalais and Gilbert, 2008). Policies and measures have been im-plemented to reduce the input of nutrients to coastal ecosystems(Duarte et al., in press) and in the Baltic Sea as well (HELCOM, 2007).

The coastal waters around Denmark were one of the first areas todocument significant decreases in nutrient concentrations on a largeregional scale resulting from an active management strategy to reducenutrients from both diffuse and point sources (Carstensen et al., 2006).Our analysis of Baltic Sea coastal ecosystems shows that many countriesin the watershed have reduced their nutrient loads resulting in lowernutrient concentrations in coastal waters (Fig. 1). However, our resultsalso show that improvements in bottom water dissolved oxygen con-centrations are not always observed (Fig. 1). From the 33 evaluatedsites only 11 show an increasing trend in dissolved oxygen of bottomwaters with time and only the Stockholm Archipelago had consistenttrends. However, the vastmajority of sites continue toworsen, especiallyalong the Danish and Finnish coasts (Figs. 1, 3 and 5), in spite of remedi-ation efforts to reduce nutrients. From a larger perspective, only a smallpercentage (~4%) of 400 sites globally suffering hypoxia have showntrends of increasing bottom water oxygen concentrations (Diaz andRosenberg, 2008).

The recovery of coastal and estuarine ecosystems also takes place ina context of global climate change (Duarte et al., 2009) with dissolvedoxygen in oceanic oxygen minimum zones decreasing due to changesassociated with global warming (Stramma et al., 2008). Warmingplays an important role by reducing the solubility of oxygen and in-creasing respiration. It also alters wind patterns, rainfall and runoff

affecting hydrological processes. Temperature increases stratificationand a stronger pycnocline reduces ventilation through vertical mixingpromoting bottomwater oxygen depletion (Hagy et al., 2004). Increasesin bottom water temperature were observed in the Finnish Archipelagoand along the Danish coast (Figs. 1, 3 and 5) where increases in AOUwere also evident. In addition, coastal ecosystems are more vulnerableto nutrient loads with temperature increases (Justić et al., 1996; Kempet al., 2005) as seen by us in the Bothnian Bay (Table 2, Fig. 2). Warmingcan induce changes in periodical biological processes such as the timingof the spring bloom and overall productivity (Nixon, 2009).

Wind speed and wind direction affect the efficiency of mixingbetween surface and bottom waters, thus changing the ventilation ofbottomwaters. Relationships between temperature increases, decreasesin wind speed and increases in oxygen demand have been observed inthe Narragansett Bay and Long Island Sound (Nixon, 2009) and Danishcoastal waters (Conley et al., 2007). In addition, climate change willalso modify rainfall patterns and snow melt generally increasing waterdischarges from land (freshwater, pollutants and nutrients) into Balticcoastal ecosystems (Meier et al., 2012). Although climate change willaffect the wind field (IPCC, 2007), we were not able to evaluate thiseffect on trends in bottom water dissolved oxygen.

4.2. Differences between systems

The recovery of bottom water oxygen concentrations followingnutrient reduction measures is rare with increases in hypoxia continu-ing to be observed globally (Conley et al., 2011; Diaz and Rosenberg,2008; Gilbert et al., 2010). Our regional evaluation of coastal ecosystemsin the Baltic Sea has shown that different regions are responding sepa-rately to reductions in nutrients. Area-specific ecological responses tocoastal eutrophication have been previously observed in the Baltic Sea(Carstensen et al., 2011; Rönnberg and Bonsdorff, 2004). In the concep-tual model of the coastal eutrophication problem by Cloern (2001) it isexplicitly recognized that system-specific attributes modulate theresponses to nutrient enrichment, leading to large differences amongcoastal systems in their sensitivity. Our response parameter used isbottom water hypoxia, whereas there are a complex suite of directand indirect responses to coastal eutrophication including changesin water transparency, distribution of bottom vegetation, in nutrientcyclingprocesses, phytoplankton community composition, and changesin ecosystem functions. The result is a wide range of estuary-specificresponses to the effects of nutrient loading on coastal eutrophicationand hypoxia (Cloern, 2001).

The coastal zone of the Baltic has a complex topography withnumerous islands and many basins separated by shallow sills resultingin many distinct isolated basins. The coastline of the Western GotlandBasin, including the Stockholm Archipelago, and Gulf of Finland, includ-ing the Finnish Archipelago, is more prone to hypoxia than other partsof the more unrestricted shoreline (Conley et al., 2011). These systemsbecome thermally stratified during the summer months. Because thecomplex topography reduces circulation and increases the residencetime of bottom waters, these ecosystems are consequently more sensi-tive to enhanced nutrient inputs from land and regularly experiencehypoxia. Our results confirm these patterns.

Stratification is a precursor for hypoxia to develop, but even perma-nently stratified coastal systems can remain normoxic provided that thebottom waters are sufficiently exchanged through bottomwater venti-lation. Hypoxia may also develop in coastal systems that stratify forshort periods only, provided that the bottom layer is so thin that respi-ratory processes, mainly from the sediments, consume all oxygenduring such short-lived events. Our results show that it is difficult toencapsulate the complex and highly dynamical physical processesgoverning oxygen supply by means of monitoring data. Such processescan be describedmore simply in open water systems, where changes inoxygen conditions are much slower (Conley et al., 2007; Carstensenet al., 2014).

198 A.M. Caballero-Alfonso et al. / Journal of Marine Systems 141 (2015) 190–199

Oxygen depletion at a sitemight be imported by the advective trans-port of saltier, low oxygen bottom water from an adjacent body; thiscould be the case of what has been observed in the present studyat Horsens Fjord (Danish Straits) and Kanholmsfjärden (StockholmArchipelago). These intrusions are characterized by nutrient-rich low-oxygen waters coming from the Belt Sea (Danish Straits) or from thebottom waters of the Baltic Proper.

4.3. Potential regime shifts

Bottom water hypoxia is not only a particularly severe disturbanceon ecosystems because it causes death of biota, changes in the habitatof living resources, but also the biogeochemical processes that controlnutrient concentrations in the water column (Conley et al., 2009a). Nu-merous studies in other coastal marine ecosystems have demonstratedthat repeated hypoxic events can help to sustain future hypoxic condi-tions (Hagy et al., 2004; Kemp et al., 2005, 2009; Turner et al., 2008).These changes suggest that a regime shift occurs in coastal marine eco-systems affected by large-scale hypoxia (Conley et al., 2007). Regimeshifts are often rapid transitions that occur in an ecosystem state withthe recovery of ecosystems delayed by internal feed-back mechanisms(Scheffer et al., 2001). The recovery of ecosystems from hypoxia mayoccur only after long periods of time (Diaz, 2001; Galloway et al.,2008; Kemp et al., 2009) or with further reductions in nutrient inputs.Perhaps the slow response of the coastal systems in the Baltic Sea to re-duced nutrient inputs is due to feed-back mechanisms such as en-hanced P release (Conley et al., 2009a, 2009b) and decreases indenitrification from sediments (Testa and Kemp, 2012) that keep eco-systems hypoxic. In addition, it may also take time for nutrients thathave accumulated in the sediments during the years of excessive nutri-ent loading to be buried out of the reactive zone or flushed out of theecosystem.

Previous studies have shown that a trend reversal is possible whenreducing nutrient loads (Carstensen et al., 2006, 2011; Nixon, 2009;Norkko et al., 2012). What is clear from our results is that there aremany coastal systemswith decreasing nutrients, but notmany showingdecreasing hypoxia. However, it is probable that the ecosystem charac-teristics, composition, components and biogeochemical processes havechanged.Most coastal marine ecosystems are impacted bymultiple fac-tors that act to stress the ecosystem (Halpern et al., 2008; Micheli et al.,2013). The removal of one pressure alone, such as nutrient reduction,may prove insufficient for ecosystem recovery when other pressures(e.g., overfishing, dredging, climate change, or pollution) have notbeen removed (Duarte et al., in press). The classification of anthropo-genic pressures shows that the highest cumulative impacts over theBaltic Sea area were in the southern and south-western areas and inthe Gulf of Finland where hypoxia is still increasing and the lowestindex values found in the Gulf of Bothnia (Korpinen et al., 2012).

4.4. Implications for ecosystem management

Reductions in nutrient loadings to coastalmarine ecosystems are theonly realistic measure to improve ecosystem health and reduce hypoxia(Conley et al., 2009b). However, confident predictions of recovery tar-gets with nutrient reductions are problematic due to the unique behav-ior of ecosystems during the recovery process (Carstensen et al., 2011).Unexpected hypoxia mitigation is proposed to have occurred withthe spread of an invasive species, e.g. the worm Marenzelleria spp.(Norkko et al., 2012). Their bioirrigation activities acted as a driver ofecological change affecting sediment P dynamics in the StockholmArchipelago, oxygenating the sediments and reducing nutrient leakagefacilitating the recovery of hypoxia. Other currently unidentified nonlin-ear interactions and feedbacks may play a positive role in ecosystemremediation if nutrient reductions are continued.

5. Conclusion

The important lesson learned is that the recovery of coastal marineecosystems from hypoxia with nutrient reductions is indeed possible(Conley et al., 2011; Diaz and Rosenberg, 2008; this study). However,nutrient reductions are not always enough and restoration effortsshould include actions to catalyze recovery and reduce the thresholdsto recovery (Duarte et al., in press). In addition, despite the multiplestressors driving hypoxia, especially the global rise in temperatures,our results demonstrate that managing nutrients can create positivefeedbacks for oxygen recovery to occur. In the absence of nutrientreductions, there will not be recovery from hypoxia in coastal marineecosystems.

Acknowledgments

This paper is a contribution from the Multistressors Project fundedby a FORMAS Strong Research Environment and the COCOA Projectfunded under the BONUS program.

References

Andersson, L., Rydberg, L., 1988. Trends in nutrient and oxygen conditions within theKattegat: effects of local nutrient supply. Estuar. Coast. Shelf Sci. 26, 559–579.

Bianchi, T.S., DiMarco, S.F., Cowan Jr., J.H., Hetland, R.D., Chapman, P., Day, J.W., Allison,M.A., 2010. The science of hypoxia in the Northern Gulf of Mexico: a review. Sci.Total Environ. 408, 1471–1484.

Carstensen, J., Conley, D.J., Andersen, J.H., Ærtebjerg, G., 2006. Coastal eutrophication andtrend reversal: a Danish case study. Limnol. Oceanogr. 51, 398–408.

Carstensen, J., Sánchez-Camacho, M., Duarte, C.M., Krause-Jensen, D., Marbà, N., 2011.Connecting the dots: responses of coastal ecosystems to changing nutrient concen-trations. Environ. Sci. Technol. 45, 9122–9132.

Carstensen, J., Andersen, J.H., Gustafsson, B.G., Conley, D.J., 2014. Deoxygenation of theBaltic Sea during the last century. Proc. Natl. Acad. Sci. 111, vol. 15, 5628–5633.http://dx.doi.org/10.1073/pnas.1323156111.

Cloern, J.E., 2001. Our evolving conceptual model of the coastal eutrophication problem.Mar. Ecol. Prog. Ser. 210, 223–253.

Conley, D.J., Carstensen, J., Ærtebjerg, G., Christensen, P.B., Dalsgaard, W., Hansen, J.L.S.,Josefson, A.B., 2007. Long-term changes and impacts of hypoxia in Danish coastalwaters. Ecol. Appl. 17, S165–S184.

Conley, D.J., Björck, S., Bonsdorff, E., Carstensen, J., Destouni, G., Gustafsson, B., Hiatanen, S.,Kortekaas, K., Huosa, H., Meier, H.E.M., Müller-Karulis, B., Nordberg, K., Norkko, A.,Nürnberg, G., Pitkänen, H., Rabalais, N.N., Rosenberg, R., Savchuk, O.P., Slomp, C.P.,Voss, M., Wulff, F., Zillén, L., 2009a. Hypoxia-related processes in the Baltic Sea. Envi-ron. Sci. Technol. 43, 3412–3420.

Conley, D.J., Bonsforff, E., Carstensen, J., Destouni, G., Gustafsson, B.G., Rabalais, N., Voss,M., Zillén, L., 2009b. Tacking hypoxia in the Baltic Sea: is engineering a solution?Environ. Sci. Technol. 43, 3407–3411.

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., RodriguezMedina, M., Lysiak-Pastuszak, E., Remeikaite-Nikiene, N., Walve, J., Wilhelms, S.,Zillén, L., 2011. Hypoxia is increasing in the coastal zone of the Baltic Sea. Environ.Sci. Technol. 45, 6777–6783.

Diaz, R.J., 2001. Overview of hypoxia around the world. J. Environ. Qual. 30, 275–281.Diaz, R.J., Rosenberg, R., 2008. Spreading dead zones and consequences for marine ecosys-

tems. Science 321, 926–929.Duarte, C.M., Conley, D.J., Carstensen, J., Sánchez-Camacho, M., 2009. Return to Neverland:

shifting baselines affect eutrophication restoration targets. Estuar. Coasts 32, 29–36.Duarte, C.M., Borja, A., Carstensen, J., Elliott, M., Krause-Jensen, D., Marbà, N., 2014. Para-

digms in the recovery of estuarine and coastal ecosystems. Estuar. Coasts. http://dx.doi.org/10.1007/s12237-013-9750-9.

Galloway, J.N., Townsend, A.R., Erisman, J.W., Bekunda, M., Cai, Z., Freney, J.R., Martinelli, L.A.,Seitzinger, S.P., Sutton, M.A., 2008. Transformation of the nitrogen cycle: recent trends,questions, and potential solutions. Science 320, 889–892.

Gilbert, D., Rabalais, N.N., Diaz, R.J., Zhang, J., 2010. Evidence for greater oxygen declinerates in the coastal ocean then in the open ocean. Biogeosciences 7, 2283–2296.

Gustafsson, B.G., Schenk, F., Blenckner, T., Eilola, K., Meier, H.E.M., Müller-Karulis, B.,Neumann, T., Ruoho-Airola, T., Savchik, O.P., Zorita, E., 2012. Reconstructing thedevelopment of Baltic Sea eutrophication 1850–2006. Ambio 41, 534–548.

Hagy, J.D., Boyton, W.R., Keefe, C.W., Wood, K.V., 2004. Hypoxia in Chesapeake Bay,1950–2001: long-term change in relation to nutrient loading and river flow. Estuaries27, 634–658.

Halpern, B.S., McLeod, K.L., Rosenberg, A.A., Crowder, L.B., 2008. Managing for cumulativeimpacts in ecosystem-based management through ocean zoning. Ocean Coast.Manag. 51, 203–211.

HELCOM, 2007. The Baltic Sea action plan; Helsinki, Finland. www.helcom.fi/BSAP/.Hetland, R.D., Di Marco, S.F., 2008. How does the character of oxygen demand control

the structure of hypoxia in the Texas–Louisiana continental shelf? J. Mar. Syst. 70,49–62.

199A.M. Caballero-Alfonso et al. / Journal of Marine Systems 141 (2015) 190–199

IPCC (Intergovernmental Panel on Climate Change, Climate Change), 2007. The PhysicalScience Basis, Contribution of Working Group I to the Fourth Assessment Report ofthe Intergovernmental Panel on Climate Change. Cambridge Univ. Press, New York.

Justić, D., Rabalais, N.N., Turner, R.E., 1996. Effects of climate change on hypoxia in coastalwaters: a doubled CO2 scenario for the northern Gulf of Mexico. Limnol. Oceanogr.41, 992–1003.

Karlson, K., Bonsdorff, E., Rosenberg, R., 2007. The impact of benthic macrofauna fornutrient fluxes from Baltic Sea sediments. Ambio 36, 161–167.

Kemp,W.M., Boynton,W.R., Adoli, J.E., Boesch, D.F., Coicourt,W.C., Brush, G., Cornwell, J.C.,Fisher, T.R., Gilbert, P.M., Hagy, J.D., Harding, L.W., Houde, E.D., Kimmel, D.G., Miller,W.D., Newell, R.I.E., Roman, M.R., Smith, E.M., Stevenson, J.C., 2005. Eutrophicationof Chesapeake Bay: historical trends and ecological interactions. Mar. Ecol. Prog. Ser.303, 1–29.

Kemp, W.M., Testa, J.M., Conley, D.J., Gilbert, D., Hagy, J.D., 2009. Temporal responsesof coastal hypoxia to nutrient loading and physical controls. Biogeosciences 6,2985–3008.

Korpinen, S., Meski, L., Andersen, J.H., Laamanen, M., 2012. Human pressures and theirpotential impact on the Baltic Sea ecosystem. Ecol. Indic. 15, 105–114.

Limburg, K.E., Olson, C., Walther, Y., Dale, D., Slomp, C.P., Hoie, H., 2011. Tracking Baltichypoxia and cod migration over millennia with natural tags. Proc. Natl. Acad. Sci.108, E177–E182.

Meier, H.E.M., Andersson, H.C., Arheimer, B., Bleckner, T., Chubarenko, B., Donnelly, C.,Eilola, K., Gustafsson, B.G., Hansson, A., Jonathan, H., Höglund, A., Kuznetsov, I.,MacKenzie, B.R., Müller-Karulis, B., Neumann, T., Niiranen, S., Piwowarczyk, J.,Raudsepp, U., Reckermann, M., Ruoho.Airola, T., Savchuk, O.P., Schenk, F., Schimanke,S., Väli, G., Weslawski, J.-M., Zorita, E., 2012. Comparing reconstructed past variationsand future projections of the Baltic Sea ecosystem-first results from multi-modelensemble simulations. Environ. Res. Lett. 7, 1–8.

Micheli, F., Halpern, B.S., Walbridge, S., Ciriaco, S., Ferretti, F., Fraschetti, S., Lewison, R.,Nykjaer, L., Rosenberg, A.A., 2013. Cumulative human impacts on Mediterraneanand Black Sea marine ecosystems: assessing current pressures and opportunities.PLoS One. http://dx.doi.org/10.1371/journal.pone.007988.

Nixon, S.W., 1995. Coastal marine eutrophication: a definition, social causes, and futureconcerns. Ophelia 41, 199–219.

Nixon, S.W., 2009. Eutrophication and the macroscope. Hydrobiologia 629, 5–19. http://dx.doi.org/10.1007/s10750-009-9759-z.

Norkko, J., Reed, D.C., Timmermann, K., Norkko, A., Gustafsson, B.G., Slomp, C.P.,Carstensen, J., Conley, D.J., 2012. A welcome can of worms? Hypoxia mitigation byan invasive species. Global Change Biol. http://dx.doi.org/10.1111/j.1365-2486.2011.02513.x.

Omstedt, A., Meuller, A., Nyberg, L., 1997. Interannual, seasonal and regional variations ofprecipitation and evaporation over the Baltic Sea. Ambio 26, 484–492.

Perus, J., Bonsdorff, E., 2004. Long-term changes in macrozoobenthos in the Ålandarchipelago northern Baltic Sea. J. Sea Res. 52, 45–56.

Rabalais, N.N., Gilbert, D., 2008. Distribution and consequences of hypoxia. In: Urban, E.,Sundby, B., Malanotte-Rizzoli, P., Melillo, J.M. (Eds.), Watersheds, Bays and BoundedSeas. Island Press, Washington, DC, pp. 209–226.

Rabalais, N.N., Diaz, R.J., Levin, L.A., Turner, R.E., Gilbert, D., Zhang, J., 2010. Dynamics anddistribution of natural and human-caused hypoxia. Biogeosciences 7, 585–619.

Rabalais, N.N., Cai, W.-J., Carstensen, J., Conley, D.J., Fry, B., Hu, X., Quinones-Rivera, Z.,Rosenberg, R., Slomp, C.P., Turner, R.E., Voss, M., Wissel, B., Zhang, J., 2014.Eutrophication-driven deoxygenation in the coastal ocean. Oceanography 27, 66–77.

Rönnberg, C., Bonsdorff, E., 2004. Baltic Sea eutrophication: area-specific ecological conse-quences. Hydrobiologia 514, 227–241.

Scheffer, M., Carpenter, S., Foley, J.A., Folke, C., Walker, B., 2001. Catastrophic shifts in eco-systems. Nature 413, 591–596.

Sohlenius, G., Emeis, K.-C., Andrén, E., Andrén, T., Kohly, A., 2001. Development of anoxiaduring the Holocene fresh-brackish water transition in the Baltic Sea. Mar. Geol. 177,221–242.

Steckbauer, A., Duarte, C.M., Carstensen, J., Vaquer-Sunyer, R., Conley, D.J., 2011. Ecosys-tem impacts of hypoxia: thresholds of hypoxia and pathways to recovery. Environ.Res. Lett. 6. http://dx.doi.org/10.1088/1748-9326/6/2/025003.

Stramma, L., Johnson, G.C., Sprintall, J., Mohrholz, V., 2008. Expanding oxygen-minimumzones in the tropical oceans. Science 320, 655–658.

Testa, J.M., Kemp,W.M., 2012. Hypoxia-induced shifts in nitrogen and phosphorus cyclingin Chesapeake Bay. Limnol. Oceanogr. 57, 835–850.

Turner, R.E., Rabalais, N.N., Justić, D., 2008. Gulf of Mexico hypoxia: alternate states and alegacy. Environ. Sci. Technol. 42, 2323–2327.

Vallin, L., Nissling, A., Westin, L., 1999. Potential factors influencing reproductive successof Baltic cod, Gadus morhua: a review. Ambio 28, 92–99.

Vaquer-Sunyer, R., Duarte, C.M., 2010. Sulfide exposure accelerates hypoxia-driven mor-tality. Limnol. Oceanogr. 55, 1075–1082.

Wulff, F., Stigebrandt, A., Rahm, L., 1990. Nutrient dynamics of the Baltic Sea. Ambio 19,126–133.

Zillén, L., Conley, D.J., Andrén, T., Andrén, E., Björck, S., 2008. Past occurrence of hypoxia inthe Baltic Sea and the role of climate variability, environmental change and humanimpact. Earth Sci. Rev. 91, 77–92.