E x t e r n a l G r o w t h C o n t r o l o f B a l t i c S e a C y a n o b a c t e r i a

Anna Zakrisson

External growth-control of Baltic Sea

cyanobacteria

Anna Zakrisson

©Anna Zakrisson, Stockholm University 2015

ISBN 978-91-7649-086-0

Printed in Sweden by Publit, Stockholm 2015

Distributor: Department of Ecology, Environment and Plant Sciences,

Stockholm University, Stockholm, Sweden.

“Science is like a love affair with nature; an elusive, tantalizing mistress.

It has all the turbulence, twists and turns of romantic love, but that's part of

the game.”

- Vilayanur S. Ramachandran

Für Luzie, Ben und Wåfflan. Dieses Buch ist für euch.

vii

Abstract

The overall aim of the study was to provide better insights to the ecological

role and impact of cyanobacteria in Baltic Sea (BS) bay, coastal and open sea

areas. Biomass and heterocyst development of diazotrophic, heterocystous

cyanobacteria and physical parameters such as nutrients and temperature

were monitored simultaneously over several years. Nitrogen fixation as well

as the transfer of fixed N within the BS food web was estimated. Still, after

decades of debate there is controversy whether eutrophication of lakes and

estuaries/coastal areas should be managed by reducing phosphorus only or

also nitrogen (Conley et al. 2009; Howarth and Paerl 2008; Schindler and

Hecky 2008; Schindler et al. 2008). Central to this debate is whether nitrogen

fixation by cyanobacteria can replace shortages of combined nitrogen (includ-

ing organic and inorganic forms of nitrogen but excluding molecular dinitro-

gen) quickly enough to make phosphorus the limiting nutrient, making nitrogen

removal pointless or even harmful (Lewis and Wurtsbaugh 2008; Paterson et

al. 2011; Scott and McCarthy 2010). It is not clear if available combined nitro-

gen inhibits heterocystous cyanobacterial nitrogen fixation (Sohm et al. 2011)

and if it is used for their growth in situ (Mulholland et al. 2004). A large eco-

system-wide experiment started in Himmerfjärden bay in 1997, where the N-

loadings from a modern sewage treatment plant (STP), located in the inner

part of the bay, were modulated. The STP creates a steep gradient of nutrients

and stable nitrogen isotopes, which can be used to study uptake of combined

nitrogen, as well as biomass development and primary productivity. The mod-

ulation of the effluent release, included years with no N-removal (2007-2008),

medium removal (2001-2002, 2005-2006, 2009-2011) and full removal (2000,

2003-2004). In addition, in 2009-2010, the effluent release depth was elevated

from 25 m depth, below the seasonal pycnocline, to 10 m depth injecting di-

rectly into the mixed layer. A 35-year long data series was used to achieve

good insights to the phytoplankton development and primary productivity in

the Baltic Sea. These in vivo long time series of monitoring data in combina-

tion with shorter time series (2-3 seasons, including measurements of colony

stoichiometry and stable isotopes), have resulted in a unique meta-dataset,

allowing for high-resolution observations on the role of cyanobacteria in the

Baltic Sea ecosystem.

viii

The greater aim of this thesis is to provide a better understanding of the

wonderful and unique Baltic Sea ecosystem with the hope of giving it a

brighter future.

ix

Contents

Abstract ................................................................................................................................. VII

Abbreviations and definitions ............................................................................................... XI

List of papers......................................................................................................................... XII

Background ............................................................................................................................ 14

Geology and geography ........................................................................................................... 14

Eutrophication and nutrient turnover ......................................................................................... 15

Phytoplankton blooms .............................................................................................................. 16

Bloom definition .................................................................................................................. 16

Spring and summer blooms ................................................................................................ 16

Baltic Sea cyanobacteria .......................................................................................................... 17

Species .............................................................................................................................. 17

Life cycle ............................................................................................................................ 18

Nitrogen fixation ....................................................................................................................... 19

Costs and benefits .............................................................................................................. 19

The heterocyst: a differentiated prokaryote cell................................................................... 20

Nitrogenase: variations in responses to combined nitrogen species ................................... 20

Stable isotopes......................................................................................................................... 21

Nitrogen stable isotopes ..................................................................................................... 22

Grazing and species interactions .............................................................................................. 22

Cyanobacteria and BS primary productivity .............................................................................. 23

Study description ................................................................................................................... 24

The Himmerfjärden eutrophication study .................................................................................. 24

Measured variables .................................................................................................................. 24

Overview of principal methods used .................................................................................... 25

Study area ................................................................................................................................ 25

Water chemistry ....................................................................................................................... 25

Cyanobacterial elemental composition ..................................................................................... 25

Biological parameters ............................................................................................................... 26

Cyanobacteria .................................................................................................................... 26

Zooplankton ....................................................................................................................... 26

Chlorophyll a ...................................................................................................................... 26

Heterocysts counts ............................................................................................................. 26

Estimates of nitrogen fixation ................................................................................................... 27

TN method ....................................................................................................................... 27

Biomass method................................................................................................................. 27

Stable isotopes and mixing models .......................................................................................... 27

Isotope analyses ................................................................................................................ 28

Mixing model: transfer of diazotrophic N to seston (< 10 m) ............................................. 28

Mixing model: transfer of diazotrophic N to zooplankton ..................................................... 28

Statistics .................................................................................................................................. 29

Hypotheses ............................................................................................................................. 30

Central findings ...................................................................................................................... 31

Hypothesis A (Paper I) ............................................................................................................. 31

Hypothesis B (Paper I) ............................................................................................................. 32

x

Hypothesis C (Paper II) ............................................................................................................ 32

Hypothesis D (Manuscript III) ................................................................................................... 34

Hypothesis E (Manuscript IV) ................................................................................................... 35

Hypothesis F (Manuscript IV) ................................................................................................... 37

Acknowledgements and Funding.......................................................................................... 40

References .............................................................................................................................. 42

xi

Abbreviations and Definitions

AIC Akaike information criterion

BS Baltic Sea

BSP Baltic Sea proper

Combined ni-

trogen

NOx [NO3+NO2] + NH4 + organic N-forms

df Degrees of freedom

DIN Dissolved inorganic nitrogen: NOx [NO3+NO2] + NH4

DIP Dissolved inorganic phosphorus: PO4

HAB Harmful algal bloom

MIB Major Baltic inflow

N Nitrogen

N2 Molecular dinitrogen

ns Not significant

P Phosphorus

PP Primary production

STP Sewage Treatment Plant

TN Total nitrogen: all N-forms particulate and dissolved

TP Total phosphorus: all P-forms particulate and dissolved

xii

List of Papers

Paper I (published)

Zakrisson, A., U. Larsson, and H. Höglander (2014) Do Baltic Sea diazo-

trophic cyanobacteria take up combined nitrogen in situ? J. Plankton Res.

(September/October 2014) 36 (5): 1368-1380

doi:10.1093/plankt/fbu053.

My contributions: Data collection in 2007 and 2009. Dr. Helena Höglander

sampled summer of 2008. Analysis of cyanobacterial and seston samples

summers of 2007-9, statistical analysis and data handling (in collaboration

with Dr. Ulf Larsson), water isotope analysis in spring 2010 (in collaboration

with the Delta Facility, Stockholm University), DAPI staining and wrote the first

draft of the manuscript.

Paper II (published)

Zakrisson, A., and U. Larsson, (2014) Regulation of heterocyst frequency in

Baltic Sea Aphanizomenon sp. J. Plankton Res. (September/October 2014)

36 (5): 1357-1367

doi:10.1093/plankt/fbu055).

My contributions: Data collection: cyanobacteria and seston in 2007 and 2009.

Dr. Helena Höglander sampled in the summer of 2008 due to parental leave.

Analysis of cyanobacterial and seston samples 2007-9 and 2011, data han-

dling and analysis (in collaboration with Dr. Ulf Larsson), heterocyst counts for

all years and wrote the first draft of the manuscript.

Paper III (submitted)

Höglander, H., A. Zakrisson, U. Larsson.

Title: Fixed nitrogen transfer in a Baltic Sea coastal area

My contributions: Data analysis, plotting and later manuscript versions.

xiii

Paper IV (manuscript)

Larsson, U., S. Nyberg, A. Zakrisson, S. Hajdu, C. Rolff, J. Walve and R.

Elmgren

Title: Long-term ecology of Baltic Sea phytoplankton: Influence of nutrients

and temperature on major groups and production

My contributions: Data analysis, plotting and co-wrote later manuscript ver-

sions.

The papers and manuscripts are referred to in the text with their roman nu-

merals. The published work is reprinted with the kind permission by Oxford

Journals©.

14

Background

Geology and geography

The Baltic Sea (BS) formed around 8000 years ago when the retreat of the

inland ice allowed salty Atlantic water to flow into the Ancylus/Mastogloia Lake

to form a large brackish basin (Rydén et al. 2003). Today the salinity is still

low, ranging from 2-3 in the Bothnian Bay to ~20 in northern Kattegat (Fig. 1)

(Granéli et al. 1990). It has an area of 377 000 km2, making it one of the largest

brackish water bodies in the world. It is very shallow, with an average depth

of 59 m and a maximum depth of 459 m at the Landsort Deep (station BY31).

Due to the fresh water input, there is usually a stable halocline at 60-80 m

depth and in summer, a seasonal pycnocline forms at 8-20 m depth in the

northern BS (Walve 2002). The water exchange between the BS and the

North Sea is mainly driven by wind but also density differences and differences

in sea level between the Danish straits and the BSP (Krauss and Brügge

1991). Despite frequent inflows of moderately saline, oxygenated water, the

density differences are rarely enough to exchange the deep water in the ba-

sins, resulting in long phases of stagnation. For example, there was for exam-

ple no deep water exchange in the Gotland Deep between 1977 and 1986

(Brettar and Rheinheimer 1991). Due to this long deep-water retention time,

oxygen is frequently exhausted with extended anoxic/hypoxic as a result

(Hansson and Andersson 2013; Schinke and Matthaus 1998). The brackish

water results in a species-poor system due to the salinity stress exerted on

both marine and limnetic species, and despite it making the system unique

and interesting, it also makes it sensitive to disturbances (Elmgren and Hill

1997).

15

Figure 1: Map of the Baltic Sea with Himmerfjärden bay (inlay). Printed with the permission of

Journal of Plankton Research (Zakrisson et al. 2014).

Eutrophication and nutrient turnover

Bloom forming cyanobacteria have been, and still are, a natural, and important

part of the Baltic Sea ecosystem. Sediment core analysis show that cyano-

bacterial blooms have occurred as far back as 7000 B.P., prior to anthropo-

genic influences (Bianchi et al. 2000). These blooms have added nitrogen to

the naturally oligotrophic Baltic Sea, supporting ecosystem production (Lars-

son et al. 2001). However, over the last century, anthropogenic activities have

rapidly eutrophied the Baltic Sea (Gustafsson et al. 2012; Larsson et al. 1985)

with peak nutrients loads in the 1980ies and thereafter with moderately de-

creased loads (Gustafsson et al. 2012). Due to the large decadal variance in

biomass, there has been some uncertainty whether there indeed has been an

increase in the open sea cyanobacterial abundance that can be related to the

increased anthropogenic phosphorus load (Kahru and Elmgren 2014; Karhu

16

et al. 1994). Albeit, the blooms occur earlier today than during previous dec-

ades (Kahru and Elmgren 2014) possibly related to increased water tempera-

tures (Manuscript IV). Most of the Baltic Sea bloom-forming cyanobacteria

species are diazotrophs; meaning that they fix gaseous nitrogen (N2). Diazo-

trophic nitrogen is estimated to support 30-90 % of the pelagic net production

in Baltic Sea proper (Larsson et al. 2001)(Manuscript IV) and to add an equiv-

alent to the annual riverine nitrogen input of ~ 480 Gg N year-1 (Gulf of Finland

excluded) (Larsson et al. 2001).

Phytoplankton blooms

Bloom definition

There are many definitions of an algal bloom. “A Dictionary of Geography©”

(Flores and Herrero 1994; Mayhew 2004) defines an algal bloom as “a rapid

increase in the population of algae in an aquatic system." This is a very broad

statement, which does not include harmfulness, the bloom biomass size, pro-

portional increase of the biomass or other coexisting factors that may render

the biomass low, even though the species is in bloom, such as ingestion by

secondary consumers. The term Harmful Algal Blooms (HABs) has historically

been used to describe large scale toxic blooms such as red tides, but is often

used for other species such as toxic cyanobacteria (Smayda 1997) for exam-

ple Nodularia spumigena, one of the key species in this study (Carmichael

1992). Wasmund (1997) defines a BS cyanobacterial bloom as > 22 µg C L-1

due to the visibility of the bloom to the naked eye. However, this excludes

Dolichospermum spp. which often occurs at maximum biomasses of ~2-4 µg

C L-1 (Zakrisson et al. 2014). Hence, it is vital to clearly define the term “bloom”,

particularly in communication with media due to the sensationalist fear-mon-

gering headlines that could result from these misunderstandings.

Spring and summer blooms

In general, the general public in the BS region associate the term “algal bloom”

with the large accumulations of smelly, decaying and toxic blooms of the cya-

nobacterium Nodularia spumigena in July-August (Hajdu et al. 2007; Sellner

et al. 2003). However, the largest BS annual bloom (based on biomass) oc-

curs in spring and consists of diatoms (Sommer 1988) and dinoflagellates

(Heiskanen and Kononen 1994), utilizing the high nutrient concentrations of

the early spring waters. The high nutrient availability is the result of high land

run-off due to the limited terrestrial plant growth and mineralization during win-

ter as well as mixing of the water column. The spring bloom also coincides

with low abundances of zooplankton, which limits grazing and allows buildup

of phytoplankton biomass. However, as temperatures increase and a thermo-

cline forms, mixed layer nitrogen (N) is depleted, and the spring diatom bloom

subsides, to be replaced by vertically migrating dinoflagellates able to utilize

17

sub pycnocline N (Höglander et al. 2004). The low post spring-bloom-water

N:P ratios result in a competitive advantage to the diazotrophic (N-fixing) fila-

mentous, heterocystous cyanobacteria (Granéli et al. 1990; Walve 2002). The

cyanobacterial bloom generally reaches a maximum in July to August and

then declines in August to October depending on wind, temperature and irra-

diances (Walve 2002; Wasmund 1997; 2001). The cyanobacterial bloom is

phosphorus limited and tends to collapse as cell molar C:P ratios reach its

maximum often > 400 (Walve and Larsson 2007).

Baltic Sea cyanobacteria

Species

The Baltic Sea ecosystem is dominated by three filamentous heterocystous

cyanobacterial species: the dominant Aphanizomenon sp., in the northern

Baltic Sea Proper often making up ~90 % of the total biomass, the toxic

Nodularia spumigena (Granéli et al. 1990; Wasmund et al. 2001) (Fig. 2), as

well as Dolichospermum spp., previously named Anabaena spp. (Wacklin et

al. 2009). N. spumigena

may form spectacular

surface accumulations and

be very patchy in its

distribution, making biomass

determinations difficult

(Almesjo and Rolff 2007).

This species favors open

sea conditions and

preferentially resides at 0-5

m depth (Hajdu et al. 2007),

in contrast to

Aphanizomenon sp., which

can be found at similar

abundances at the coast

and the open sea

(Manuscript IV) and occurs at 5-10 m depth (Hajdu et al. 2007; Walve and

Larsson 2007; Wasmund et al. 2001). Stal et al. (2003) also found that

Aphanizomenon sp. was more common in the northern BSP, whereas N.

spumigena had higher biomasses in the southern part. Filamentous

cyanobacteria contain gas vesicles, allowing them to regulate their buoyancy

(Walsby et al. 1995). N. spumigena is generally not found during the winter

months (S. Hajdu, pers. comm.) and is likely overwintering in the

cyanobacterial spore-form, called “akinete” (Adams and Duggan 1999),

whereas Aphanizomenon sp. vegetative filaments frequently can be found in

Figure 2: N. spumigena and Aphanizomenon sp.

Photo kindly provided by Dr. Helena Höglander

18

the winter waters, though at very low biovolume (Laamanen et al. 2005).

Nonetheless, akinete formation has also been observed in Baltic Sea

Aphanizomenon sp. at the coastal reference station B1 (Fig. 1) (Zakrisson,

unpubl.). Akinetes are highly decomposition resistant and show lower rates of

protein synthesis and photosynthetic activity than vegetative cells and may

remain dormant for several years (Huber 1984; Yamamoto 1995).

Furthermore, akinetes may accumulate 5-30 times higher RNA and DNA

concentrations than vegetative cells. This is a prerequisite for a fast onset and

high rate of growth. Akinetes also often contain a store of cyanophycin, which

functions as an internal nitrogen store (Huber 1984).

Life-cycle

The cyanobacterial (order Nostocales, which includes Aphanizomenon sp., N.

spumigena and Dolichospermum spp.) life-cycle is summarized in Figure 3.

Shortly; In spring, cyanobacterial recruiting cells germinate from akinetes in

response to external factors such as increased light, high phosphorus, low

N:P ratios, by forming gas vesicles causing them to ascend the water column

(Hense and Beckmann 2006). Huber (1985) showed that N. spumigena

akinete germination actually did not require any N at all, but did require phos-

phorus, similarly to the spores of the toxic, diazotrophic cyanobacteria Micro-

cystis, frequently found in freshwater systems (Ståhl-Deblanco et al. 2003).

Sometimes vegetative cells are present in a dormant state over winter (Suik-

kanen et al. 2010). When growth is initiated, and as internal nitrogen stores

are depleted, heterocyst differentiation is initiated. As light and nutrients de-

crease towards the end of summer, akinete formation may be triggered. Due

to the absence of gas vacuoles the akinetes sediment out of the photic zone.

19

Figure 3: Model life cycle of a heterocystous, filamentous cyanobacterium. Adapted from Hense

and Beckmann (2006).

Nitrogen fixation

Costs and benefits

The ability to fix N2 (diazotrophy) is a great advantage for life in an N-depleted

ocean. However, the process is energetically highly demanding, requiring 16

ATP per molecule fixed dinitrogen (Taiz and Zeiger 2002). Hence it has been

postulated that diazotrophs will shut down, or down-regulate N-fixation in the

presence of combined N (NOx [NO3+NO2] + NH4 + organic N-forms). This re-

sponse is reported from several instances, but with immense variation in the

response both between species and between strains (Sanz-Alférez and Del

Campo 1994) (Vintila and El-Shehawy 2007).

20

The heterocyst: a differentiated prokaryote cell

The cyanobacterial species included in this survey (Aphanizomenon sp., Nod-

ularia spumigena and Dolichospermum spp.), are all heterocystous, meaning

that they fix nitrogen in thick walled, terminally differentiated cells called het-

erocysts (= heterocytes) (Adams and Duggan 1999). The heterocysts lack the

oxygen-evolving photosystem II (Stephens et al. 2003), and have a high res-

piratory rate (Adams and Duggan 1999) to maintain the cell’s oxygen concen-

trations as low as possible, due to the terminal inactivation of the nitrogen-

fixing enzyme (nitrogenase) by oxygen. The high respiratory rate also pro-

vides the cell with substantial amounts of ATP to convert molecular dinitrogen

to nitrate. Nitrogenous compounds are transported out of the heterocysts in a

periplasm (Flores et al. 2006) or plasmodesmata and photosynthetic products

in the other direction (Adams and Duggan 1999). Some species and strains

may halt heterocyst differentiation in high combined nitrogen availability. An-

abaena CA (ATCC 33047), for example, completely halted heterocyst differ-

entiation when exposed to KNO3 and NH4NO3, but not when exposed to

NH4Cl. This resulted in only 35 % reduction compared to growing on molecular

dinitrogen only (Bottomley et al. 1979). However, Nostoc punctiforme, Ana-

baena sp. PCC 7120 and A. variabilis (commonly used cyanobacterial model

organisms) reduced heterocyst frequencies as a response to ammonium or

nitrate (Adams and Duggan 1999; Bothe 1982; Guerrero and Lara 1987).

Nonetheless, N. spumigena, strain AV1 did not reduce heterocyst frequencies

in response to NH4, despite a reduction in N-fixation rate (acetylene reduction

assay). Similar insensitivity was found for N. spumigena, strains M1 and M2

regarding the effect of NO3 on heterocyst frequency (Sanz-Alférez and Del

Campo 1994). Hence, a variety of responses exists and is a reminder not to

draw too general conclusions for natural occurring populations from model

systems. Not only nitrogen, but also other factors, such as low phosphorus

availability in combination with low light intensities may affect heterocyst for-

mation and resulted in a heterocyst frequency reduction in Anabaena flos-

aquae (Thompson et al. 1994). whereas elevated levels of phosphorus in a

mesocosm experiment increased heterocyst frequencies in the Baltic Sea

Aphanizomenon sp. (Lindahl et al. 1980).

Nitrogenase: variations in responses to combined nitrogen species

Heterocyst differentiation is a relatively slow process compared with the reg-

ulation of the nitrogenase enzyme. It takes about 6-12 hours for the transition

from a vegetative cell to a proheterocyst (an intermediate stage where exit

from the differentiation process still is possible) and 12-20 hours to a mature

heterocyst (using the model organisms Nostoc or Anabaena)(Adams 2000).

On the contrary, modifications of the nitrogenase enzyme are fast and will

more quickly modify N-fixation rates and are probably a better modulation

method (Hense and Beckmann 2006). Enzymatic regulation also seems to be

species or strain specific and also depends on nitrogen species, for example,

21

Baltic Sea N. spumigena (strain AV1) down-regulated N-fixation (acetylene

reduction assay) in the presence of NH4, but not of NO3 (mM concentrations)

(Vintila and El-Shehawy 2007). Additionally, N. spumigena (strain B1) did not

modify N-fixation rates in response to NO3 (Lehtimäki et al. 1997). On the

contrary, Anabaena CA (ATCC 33047) reduced N-fixation rates more strongly

in response to NO3 compared with NH4 (Bottomley et al. 1979). The non-het-

erocystous Trichodesmium erythraeum (IMS-101) reduced N-fixation rates

with 70 % (acetylene reduction assay) when exposed to 10 µM NO3 (Holl and

Montoya 2005). Not only combined nitrogen may affect N-fixation rates but

sulfate has also been found to inhibit N-fixation rates in naturally occurring

Baltic Sea populations of N. spumigena (1999). All of these strains and spe-

cies have evolved in distinct limnetic, brackish and marine niches. Since nitro-

gen-fixation is energetically costly and such a central process for their survival

and defines their competitive advantage over smaller fast-growing phytoplank-

ton, it would be sensible to assume that its regulation may be very specific to

be able to respond quickly to their particular environment. Therefore, it is of

great benefit to investigate naturally occurring populations for a better insight

to how these systems function and how they are regulated in vivo.

Stable isotopes

Stable isotopes are useful for estimating N uptake by diazotrophs and enables

tracing of N derived from various sources up the food web. Many elements

exist in the form of different isotopes, i.e. atoms with different numbers of neu-

trons. Some isotopes are unstable and decay through radioactive breakdown,

but others are not radioactive and referred to as stable isotopes. Such iso-

topes are commonly used in ecosystem studies to investigate the transfer of

nutrients between different trophic levels. A stable isotope is an atom that has

acquired extra neutrons, but with a stable nucleus, with as many, or slightly

more neutrons than protons (Fry 2006). Differences in the number of neutrons

usually do not affect the chemistry of the atom as the electron sphere remains

identical irrespective of the neutron number (Fry 2006). Though, as the iso-

tope with more neutrons is heavier, the lighter isotope will be reacting faster

than its heavy counterpart, resulting in fractionation (Fry 2006), enabling

trophic studies like the one presented here. The stable isotope systems used

here are 15N/14N and 13C/12C. The -notation has been used.

(‰) = [(𝑅𝑠𝑎𝑚𝑝𝑙𝑒 − 𝑅𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑)/𝑅𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 ] × 1000

where R is the ratio of heavy to light isotope.

22

Nitrogen stable isotopes

The N-isotope standard (AIR) is based on atmospheric N and has by definition

= 0 ‰ (Fry 2006), whereas N-fixation renders values of -2.7 to 1.2 ‰ (Macko

et al. 1987; Minagawa and Wada 1984). Denitrification and anammox re-

leases light N2 to the atmosphere with a remaining enriched (in 15N) biomass,

as well as NO32-, NO2

-, NH4+ (Altabet et al. 1999).

Grazing and species-interactions

Cyanobacteria are commonly considered as poor food sources for zooplank-

ton due to their low polyunsaturated fatty acid content. For example 20:53

and 22:63 are needed for crustacean reproduction (Brett and Muller-Navarra

1997; Kozlowsky-Suzuki et al. 2007). Due to their filamentous form, cyano-

bacteria may also clog the feeding appendages of many zooplankton species,

in particular cladocerans, causing starvation and reduced growth (Lampert

1987). Nonetheless, there is great variation in sensitivity, not only between

zooplankton species, but also within species due to the different zooplankton

life-stages (Xu and Burns 1991). Many cyanobacteria produce toxins such as

the hepatotoxin nodularin, by N. spumigena (Bianchi et al. 2000; Granéli et al.

1990). Sensitivity of secondary producers varies considerably as, for example,

juvenile growing fishes are much more susceptible to toxins and deleterious

diets than adult fish (Karjalainen et al. 2007). Many secondary consumers are

able to avoid ingesting toxic cyanobacteria but they are still exposed to water

dissolved toxins (Karjalainen et al. 2003). However, not all effects of cyano-

bacteria are deleterious. The refuge hypothesis suggests that cyanobacterial

colonies may act as refuges due to increased turbidity and reduced visual

acuity of predators (Utne-Palm 2002). Also, clogging of predator feeding ap-

pendages may protect their prey (Karjalainen et al. 2007). In addition, positive

allelopathic effects of cell-free cyanobacterial filtrates (Dolichospermum sp.,

N. spumigena and Aphanizomenon sp.) have been found for several organ-

isms ranging from the cyanobacteria Snowella spp., Pseudanabaena spp.,

Dolichospermum spp. to the chlorophyte Oocystis sp., the dinoflagellate Am-

phidinium sp. as well as several nanoflagellates. The interaction between cy-

anobacteria, grazers and secondary consumers is extremely complex

(Gorokhova and Engstrom-Ost 2009) as demonstrated by reduction in toxin

production in N. spumigena when grazed by the copepod Eurytemora affinis

(Gorokhova and Engstrom-Ost 2009). Also, N. spumigena had a positive ef-

fect on growth and egg production on the copepods Acartia spp. and Eury-

temora affinis, though Aphanizomenon sp. had a negative effect (Hogfors et

al. 2014) despite that N. spumigena is toxic and Aphanizomenon sp. consid-

ered non-toxic (Carmichael 1992).

23

Cyanobacteria and Baltic Sea primary productivity

In summer cyanobacteria may contribute to up to a third of the total primary

production (Stal et al. 1999; Stal and Walsby 2000). Towards the end of the

blooms, cells lyse, and their contents are emptied into the water mass (Lindahl

et al. 1980; Schulz and Kaiser 1986), injecting nutrients to the summer mixed

layer. Cyanobacteria leak significant amounts of nitrogen, even during the log-

arithmic growth phase (Ohlendieck et al. 2000; Ploug et al. 2011; Ploug et al.

2010). Ploug et al. (2011; 2010) have estimated up to 35 % loss of fixed nitro-

gen as ammonium, supporting summer non-diazotroph primary productivity

(Ohlendieck et al. 2007).

24

Study description

The Himmerfjärden Eutrophication Study

In 1997, nitrogen removal was implemented in the sewage treatment plant

(STP) located in the inner parts of the Himmerfjärden bay (Fig. 1). Before this

implementation, a large excess of nitrogen relative to phosphorus suppressed

diazotrophic biomass in the inner parts of the bay, as small and fast growing

non-diazotroph phytoplankton out-compete the filamentous cyanobacteria for

phosphorus when nitrogen is in excess in the mixed layer. From 1997 onward,

the reduction in the nitrogen load significantly increased the diazotrophic bio-

mass. However, total Chl a decreased and an increased Secchi depth was

achieved, proving the benefits of effluent nitrogen removal.

The study includes large-scale modifications of both STP effluent nitrogen

content, as well as release depth. The experimental modifications started in

2001 with authority permission. In 1998-2000, 2003-2004 effluent nitrogen re-

moval was fully functional, 2001-2002 and 2005-2006 and 2009-2010 there

was a medium release and in, 2007 and 2008 the effluent nitrogen removal

was discontinued to investigate the potential suppression of the diazotrophic

biomass to pre-1997 levels. In 2009-2010 the effluent release was moved

from 25 m depth, below the seasonal pycnocline to 10 m depth, injecting the

effluent directly into the productive zone. The aim was to investigate if the

injection of nitrogen is sufficient to maintain low Chl a, high Secchi depth and

to reduce the diazotroph biomass.

Measured variables

Within the framework of the present study, a large variety of physical and bio-

logical variables have been measured, such as Chl a, nutrients (total nitrogen

(TN), total phosphorus (TP), dissolved inorganic nitrogen (DIN: NH4 + NOx

[NO3 + NO2]) and dissolved inorganic phosphorus (DIP)), temperature, salin-

ity, Secchi depth, biomass of phytoplankton and various zooplankton (only B1

and H4). These parameters were measured every week in spring, every sec-

ond week over the summer and autumn seasons and monthly in winter.

25

Overview of the principal methods used

Study area

Himmerfjärden bay is located approximately 60 km south of Stockholm. Its

surface area is 232 km2, with a mean depth of 17 m. Sills separate it into four

basins, extending north-to-south, with its northernmost part connecting to

Lake Mälaren through Hallsfjärden bay (Fig. 1). A modern sewage treatment

plant (STP) is located in the inner parts of the bay (Fig. 1). Prior to the change

of the STPs discharge depth from 25 m to 10 m (2009 to 2010), the estuarine

circulation transported some of the treated sewage northward when stratified

(Engqvist 1996). Water column stratification commonly develops in late April

or beginning of May (U. Larsson, unpubl.). In papers I, II and III, the sampling

was in May-September (2000 to 2001 and 2007 to 2009). Station B1 was used

as a coastal reference station, BY31, located at the Landsort Deep (459 m

depth) as an open sea reference station (Fig. 1). Both B1 and BY31 are mon-

itored by the Swedish National Environment Monitoring Program.

Water chemistry

Water sampling was carried out every two weeks at every 5 m down to 25-50

m depth depending on station depth. Total phosphorus (TP) and nitrogen (TN)

(inorganic, organic and particulate forms), were digested simultaneously using

a modified method (Koroleff 1983) starting with high and ending with low pH

and analyzed on an ALPKEM O. I. Analytical Flow Solution IV (method #

319528). Dissolved inorganic phosphate (DIP) and dissolved inorganic nitro-

gen (DIN: NH4, NOx=NO3+NO2) were measured on ALPKEM O. I. Analytical

Flow Solution IV, using methods # 319528, # 319526 and # 319527. Water

salinity and temperature in situ were measured with a CTD probe.

Cyanobacterial elemental composition

Sorted cyanobacterial colonies were placed in capsules (SÄNTIS Analysis)

for CN and P analysis on small acid washed glass plates. After drying at 60

°C for three days (Walve and Larsson 2007), P-analysis was performed on an

26

ALPKEM O. I. Analytical Flow Solution IV (method # 319528, edition 990121),

following combustion at 500 oC and persulphate digestion, while CN was an-

alysed using a Leco CHNS-932.

Biological parameters

Cyanobacteria

Cyanobacteria were sampled using a hose (inner diameter was 2.5 cm) with

a valve in the upper end and a weight in the lower, from the surface down to

20 m (stations BY31 and B1) or 14 m (stations H2, H3, H4, H5, H6, H8). By

having the valve open when lowering the tube and closing when the lower end

of the tube had reached the desired depth, an integrated sample of the water

column was obtained. Sub-samples of 200 mL were preserved with 0.8 mL

acetic Lugol’s iodine solution. Abundances were determined following the Hel-

sinki Commission (HELCOM) guidelines (HELCOM 1988). Counts were con-

ducted using an Utermöhl chamber bottom (25 ml) and a Wild M40 inverted

microscope. Species-specific mean cell volumes were multiplied by cell num-

bers to obtain biomass, assuming a density of 1 g mL-1 (Olenina et al. 2006).

Zooplankton

Samples were taken (2000 to 2001) with a 25 L water-sampler with 5 m inter-

vals (stations BY31 [0-30 m], B1 [0-35 m] and H4 [0-25 m]) prior to sieving by

a 90 µm net. A final concentration of 4 % formaldehyde was used to preserve

the samples. A Kott splitter was used to divide the zooplankton samples, that

were counted in 10 ml Utermöhl chambers using a Leitz Fluovert FS micro-

scope (80 x magnification). Biomass was calculated according to Hernroth

(1985).

Chlorophyll a

GF/F filters were used to collect phytoplankton and were ground in 90 % ace-

tone, extracted for 2h at room temperature before spectrophotometric analysis

(SIS-standard SS 02 81 46).

Heterocyst counts

Two different counting methods were used, 1) Monitoring dataset (used in Pa-

per II; Dataset1998-2011) and 2) nutrient and isotope dataset (used in Paper II;

Dataset2007-2009). For the former dataset Cyanobacteria were collected simul-

taneously as the biomass measurements (see “Biomass determinations”).

The full chamber was counted, but no minimum filament length was counted,

creating a greater uncertainty for heterocyst frequencies at lower biovolumes.

For the latter, Lugol preserved samples were collected by using vertical nets

27

(90 m) hauls (0-10 m) and were counted for at least 10 mm filaments using

a Leica DMIRB inverted microscope.

Estimates of nitrogen fixation

Two different methods were used for estimates on N-fixation to enable a com-

parison of the method and their relative merits.

TN method

Summer mixed layer TN increase (TN) was calculated subtracting the mean

of the 10 m sub pycnocline from the mean pycnocline TN concentrations

(Larsson et al. 2001). This was done in order to minimize the effect of N-rich

upwellings into the mixed layer. Pycnoclines were defined at 20 m depth at

BY31, 15 m at B1 and 10 m for the inner bay stations (H2-H6) using the Brunt-

Väisälä buoyancy frequencies (NB):

𝑁𝐵 = −(𝑔 × − 1(𝑝 × 𝑧 − 1))0.5

where = mean density, g = the gravitational acceleration and (p×z-1) is the

change in density per unit of depth. We assume that the TN increase is of

diazotrophic origin. The estimates were adjusted for atmospheric deposition

(6 mmol m-2 month-1) (Larsson et al. 2001) and for particle settling, which was

estimated from sedimentation traps at BY31 (2000).

Biomass method

Nitrogen fixation was also estimated (Paper III and IV) using biomass (Aphani-

zomenon sp. and N. spumigena) together with cell specific nitrogen-fixation

obtained from Ploug et al. (2011; 2010). Mean N-fixation of 133 x 10-15 mol

cell-1 day-1 for Aphanizomenon sp., and 264 fmol cell-1 day-1 for N. spumigena

(NH4 release included) were used (Ploug et al. 2011; Ploug et al. 2010), and

15:9 h day: night cycle. Area for estimation of basin wide N-fixation was taken

at 4 m depth (Engqvist 1996).

Stable isotopes and mixing models

Stable isotopes can be used to trace nutrients in a system due to the fraction-

ation that occurs during biological uptake (Fry 2006). The mixing models used

here take into consideration the isotopic signature as well as the concentration

of each N and/or C of each source used in the model.

28

Isotope analyses

δ15N in cyanobacteria and seston was determined by the UC DAVIS Stable

Isotope Facility using a PDZ Europa ANCA-GSL elemental analyzer inter-

faced to a PDZ Europa 20-20 isotope ratio mass spectrometer (IRMS) (Sercon

Ltd., Cheshire, UK). The per mil difference in isotopic composition () between

the samples and the standard reference material (atmospheric nitrogen) was

calculated according to Peterson and Fry (Peterson & Fry, 1987):

(‰) = [(Rsample-Rstandard)×Rstandard-1]×1000, where, R = 14N/15N

Mixing model: transfer of diazotrophic N to seston (< 10 m)

Paper I and Paper II: We used a simple one source mixing model to investi-

gate the impact of fixed N on the summer minimum in seston δ15N using a

mixing model (Fry 2006). Linear interpolations in daily steps were used for

seston, cyanobacteria N content and cyanobacterial δ15N, from two weeks be-

fore and to the minimum in seston δ15N (Zakrisson et al. 2014).

15𝑁𝑠𝑒𝑠 𝑑𝑎𝑦𝑋+1 = 15𝑁𝑐𝑦 𝑑𝑎𝑦𝑋 × 𝑓2 + 15𝑁𝑠𝑒𝑠 𝑑𝑎𝑦𝑋 × 𝑓1

where,

𝑓1 =𝑠𝑒𝑠𝑡𝑜𝑛 𝑁 𝑐𝑜𝑛𝑡𝑒𝑛𝑡

(𝑠𝑒𝑠𝑡𝑜𝑛 𝑁 𝑐𝑜𝑛𝑡𝑒𝑛𝑡 + 𝑐𝑦𝑎𝑛𝑜𝑏𝑎𝑐𝑡𝑒𝑟𝑖𝑎𝑙 𝑁 𝑐𝑜𝑛𝑡𝑒𝑛𝑡)

and

𝑓2 =𝑐𝑦𝑎𝑛𝑜𝑏𝑎𝑐𝑡𝑒𝑟𝑖𝑎𝑙 𝑁 𝑐𝑜𝑛𝑡𝑒𝑛𝑡

(𝑠𝑒𝑠𝑡𝑜𝑛 𝑁 𝑐𝑜𝑛𝑡𝑒𝑛𝑡 + 𝑐𝑦𝑎𝑛𝑜𝑏𝑎𝑐𝑡𝑒𝑟𝑖𝑎𝑙 𝑁 𝑐𝑜𝑛𝑡𝑒𝑛𝑡)

Mixing model: transfer of diazotrophic N to zooplankton

Paper III: Transfer of fixed N to zooplankton (<200 m) was calculated using

a one-source mixing model according to Fry et al. (2006).

𝑓3 =15𝑧𝑝𝑙𝑡 − 15𝑧𝑝𝑙0 − ∆

15𝑐𝑦𝑡 − 15𝑧𝑝𝑙0

and

𝑓4 = 1 − 𝑓1

Where 15𝑧𝑝𝑙𝑡 is zooplankton 15N at time t and 15𝑧𝑝𝑙0 is zooplankton 15N

prior to the cyanobacterial bloom. 15𝑐𝑦𝑡 is cyanobacterial 15N at time t. is

the fractionation factor (3.4) (Minagawa and Wada 1984). Hence, the ratio of

the fractional contributions to the total material (𝑓𝑡𝑜𝑡𝑎𝑙) is:

29

𝑓𝑡𝑜𝑡𝑎𝑙 =𝑓3 × 𝑊𝑧𝑝𝑙𝑡

𝑓3 × 𝑊𝑧𝑝𝑙𝑡 + 𝑓4 × 𝑊𝑐𝑦𝑡

Where 𝑊𝑧𝑝𝑙𝑡 is the amount N in the zooplankton biomass per liter water at

time t, assuming a conversion factor from wet to dry weight of 10 % (Dumont

et al. 1975) or 15 % (Mendoza 1993). 𝑊𝑐𝑦𝑡 is the amount of N in the cyano-

bacterial biomass per liter (g N dm-3) at time t. Thus,

𝑓𝑡𝑜𝑡𝑎𝑙𝑐𝑦 = 1 − 𝑓𝑡𝑜𝑡𝑎𝑙

Where 𝑓𝑡𝑜𝑡𝑎𝑙𝑐𝑦 is the fractional contribution of diazotrophic N to the zooplank-

ton community (>200 m). This value was then adjusted by the ratio of the

observed to expected values from the model above describing the transfer of

diazotrophic N to the seston pool since we are assuming that the dominating

cyanobacterium Aphanizomenon sp. (> 90 % of the total cyanobacterial bio-

mass) is not significantly grazed by the zooplankton (Hogfors et al. 2014) and

that the transfer of fixed N to the zooplankton pool is mainly secondary.

Statistics

We used R 2.13.2 (R Core Team 2008), RStudio Version 0.95.265 (©2009-

2011 RStudio, Inc.) for data analysis and Microsoft Excel 2003 as spread-

sheet. Spearman’s rank correlation from the coin package in R for bay gradi-

ent analyses (Hothorn et al. 2008). The Wilcoxon signed rank test was used

to compare paired stations (coin package). Nonlinear trends were analyzed

by additive modeling with the mgcv package in R (Wood 2003). Backward

stepwise elimination using Akaike’s information criterion (AIC) was used to

select the most parsimonious model (Burnham and Anderson 2004; Zuur et

al. 2009). Models with AIC differing < 3 units were considered equal, and the

least complex model was selected (Burnham and Anderson 2004). The num-

ber of knots were set to 4 to avoid over-fitting and gamma = 1.4 to inflate

degrees of freedom (df) for model selection. Collinearity was detected by vis-

ual inspection using boxplots as well as variance inflation factor (VIF) analysis

and Pearson product moment correlation coefficients (PCC). VIF>3, as well

as -0.4>PCC>0.4 were considered collinear (Zuur et al. 2009). The assump-

tions of independence, homoscedasticity and normality were all met unless

otherwise stated. The Kruskal-Wallis test was used as a non-parametric alter-

native to ANOVA. The 0.05 confidence level was used for all analyses. For

graphics, the following R-packages were used: lattice (Sarkar 2008), latticeEx-

tra (Sarkar and Andrews 2013), ggplot2 (Wickham 2009), plyr (Wickham

2011), zoo (Zeileis and Grothendieck 2005).

30

Hypotheses

We tested the following hypotheses:

A. Paper I: Uptake of combined nitrogen will reduce nitrogen fixation.

B. Paper I: Diazotrophic biomass is reduced when inorganic nitrogen

(NO3, NO2 and NH4) is injected into the mixed layer by competition

for phosphorus with non-diazotrophic species cyanobacteria (bio-

mass primarily made up by Aphanizomenon sp.).

C. Paper II: Heterocyst frequency is determined by combined nitrogen

availability.

D. Manuscript III: Diazotrophic nitrogen contributes significantly to

coastal nitrogen input and can be traced through the food web.

E. Manuscript IV: Baltic Sea spring phytoplankton primary production

is nitrogen limited.

F. Manuscript IV: Diazotrophic nitrogen boosts summer phytoplankton

primary production.

G. Manuscript IV: Filamentous diazotrophs have increased over the

past decades in both coastal and open sea areas due to changed

temperatures and nutrient concentrations.

31

Central Findings

Hypothesis A (Paper I):

“Uptake of combined nitrogen will reduce nitrogen fixation”

We found no indication of combined nitrogen uptake in Baltic Sea bulk cyano-

bacteria, mainly Aphanizomenon sp., since 15N remained similarly depleted

along the bay (late season data excluded), despite pronounced gradients in

total nitrogen (TN), dissolved inorganic nitrogen (DIN) and seston 15N. Ses-

ton 15N being a proxy for the nitrogen isotopic signal of non-diazotrophic algal

nitrogen uptake. In addition, over-wintering filaments of Aphanizomenon sp.

lacked heterocysts, but 15N remained as depleted as during maximum growth

during summer and the formation of new heterocysts in spring did not affect

15N further. This suggests that not even over wintering heterocyst free fila-

ments of Aphanizomenon sp. take up combined nitrogen. Later in the season

(mid-August), coinciding with a Dolichospermum spp. bloom, 15N did in-

crease (Fig. 4), indicative of combined nitrogen uptake in this species. We

therefore conclude that, contrary to previous belief, the larger portion of the

BS bulk cyanobacterial nitrogen fixation is not affected by ecologically relevant

combined nitrogen concentrations.

32

Figure 4: Seston and cyanobacterial nitrogen stable isotope ratios 15N (‰) in 2007 to 2009 and

at stations B1, H2, H3 and H4 (Zakrisson et al. 2014). Printed with the permission of Journal of

Plankton Research.

Hypothesis B (Paper I):

“Diazotrophic biomass is reduced when inorganic nitrogen (NO3, NO2 and

NH4) is injected into the mixed layer by competition for phosphorus with non-

diazotrophic species cyanobacteria (biomass primarily made up by

Aphanizomenon sp.).”

Cyanobacterial biomass was significantly lower at station H5, compared with

the outer stations, likely due to the higher nitrogen concentrations at this sta-

tion near the STP effluent release (Paper I: Fig. 2). The excess N increases

competition for phosphorus and trace elements, by the nitrogen limited non-

diazotroph organisms, likely reducing diazotroph growth rates (Walve and

Larsson 2007). In 2007-8 nitrogen removal in the STP was discontinued and

in 2009 resumed, but effluent release moved from 25 m depth to 10 m depth

(above the seasonal pycnocline). The treated sewage contained a large ex-

cess of nitrogen (mainly inorganic forms) compared to phosphorus, also in

2009. Despite the continued nitrogen removal in 2009, the injection of the

33

treated sewage into the mixed layer did not significantly change the diazotroph

biomass or Chl a concentrations compared with the previous two years with

discontinued nitrogen removal (Paper I: Figs. 2 & 3). This gives an indication

that it may be possible to control diazotroph biomass and at the same time

achieve low Chl a by keeping nitrogen removal in the sewage treatment plants,

but to change the effluent release depth to above the seasonal pycnocline.

Nonetheless, further conclusions have to await the ongoing replication since

variability between years can be substantial, as indicated for 2008 (Paper I:

Figs. 23).

Hypothesis C (Paper II):

“Heterocyst frequency is determined by combined nitrogen availability.”

We found that heterocyst frequency is not determined by combined nitrogen

availability. Despite the strong gradients in TN and TP (Paper II: Fig. 2), we

found no heterocyst frequency gradient along the bay, other than a higher

heterocyst frequency at H2 compared with the other stations (Paper II: Fig. 2;

F(6,753)=4.02, p <0.001, Tukey’s HSD post hoc test [May-September]). Moreo-

ver, at H5, the station with the greatest impact from the STP, heterocyst fre-

quencies did not significantly differ between the years with full and no N-re-

moval, despite almost four times higher median DIN concentrations (0.667

compared to 0.171 μM) in years without N-removal. In addition, heterocyst

formation in spring occurred four to six weeks after depletion of mixed layer

nitrogen by the spring bloom (Fig. 5), indicating that other factors may be of

more importance for initiation of the differentiation process, such as light and

temperature. We found granulate structures in over wintering heterocyst free

filaments of Aphanizomenon sp. that decreased in number as spring pro-

ceeded (6.4-6.6 granules (m fil)-1 in March to 4.4 granules (m fil)-1 in late

April 2011). However, δ15N did not change significantly over this period

(ANOVA: δ15N = -1.6 0.1 ‰), despite the significant reduction in cell nitrogen

content (F(2,7) = 12.74, p<0.01, Tukey’s HSD post hoc test), evidential of nitro-

gen storage of fixed nitrogen during the winter season, possibly as cyanophy-

cin.

The heterocyst frequency varied significantly over the season at all sta-

tions (Paper II: GAM: F(edf=2.8)=52.6, n=813, p<0.0001), with maximum values

in mid-May (as heterocysts appear) and beginning of September (Paper II:

Fig. 3). The heterocyst frequency minimum coincided with temperature maxi-

mum. Based on the 14 year long dataset (Dataset1998-2011), heterocyst fre-

quency significantly decreased with increasing temperatures. This may be in-

dicative of a more efficiently functioning nitrogenase enzyme at higher tem-

peratures (Fig. 6). We found no significant correlation between colony C:P

34

content and heterocyst frequency, though colony C:N decreased and colony

%N increased as heterocyst frequency increased. This is likely and simply due

to the fact that more heterocysts fix more nitrogen.

Figure 5: Depletion of mixed layer (upper 10 m) dissolved inorganic nitrogen (DIN) by the spring

bloom in 1998-2011. Arrows depict the first appearance of heterocysts (Paper II) (Zakrisson and

Larsson 2014). Printed with the permission of Journal of Plankton Research.

Figure 6: Relationship between heterocyst frequency and temperature at stations BY31, B1, H2,

H3, H4, H5 in 1998 to 2011 (Paper II) (Zakrisson and Larsson 2014). Printed with the permission

of Journal of Plankton Research.

Hypothesis D (Manuscript III):

“Diazotrophic nitrogen contributes significantly to coastal nitrogen input and

can be traced through the food web.”

35

Mixing models, based on natural abundances of stable nitrogen isotopes,

were used to estimate the transfer of fixed nitrogen though the food web. We

found a significant transfer of diazotrophic N to both seston and zooplankton

also in the coastal area, indicating that fixed nitrogen is an important source

of nitrogen not only in the open ocean. The transfer of fixed nitrogen from

diazotrophs to zooplankton was estimated to be ~20 % at the coastal refer-

ence station (B1) during the most productive summer weeks (Table 4). We

also estimated the actual amounts nitrogen fixed in the bay, as well as at the

coastal (B1) and open sea (BY31) reference stations based on the mixed layer

total nitrogen increase during the bloom period. Nitrogen fixation was esti-

mated to 80-90 mmol N m-2 year-1 (open sea) and ~32-102 mmol N m-2 year-1

(bay) with a higher uncertainty for the bay due frequent up-mixing of nutrient

rich sub-pycnocline water. To put this into context, summer nitrogen fixation

in the bay was 42-112 tons per year, with a median of 77 tons per year, which

is ~30 % of the mean STP nitrogen load of 170-370 tons per year.

Table 4: Proportion (%) fixed N that is transferred to the zooplankton N-pool via seston. Direct

grazing is assumed to be negligible due to the dominance (>90 %) of Aphanizomenon sp. (Hog-

fors et al. 2014). See Methods section for a description of the model. We used 15 respectively 10

% as a conversion factor from wet to dry weight for zooplankton. “Adjusted transfer” are adjusted

for the ratio of the expected and observed values from transfer calculations of diazotrophic N to

the seston pool (See Methods).

Year Week Station

Ratio of ob-

served vs.

expected

transfer of

diazotrophic

N to seston

at the bloom

peak

15 % wwt

to dwt

Prop. (%)

fixed N to

zooplankton

(FtotalCy x

100)

10 % wwt

to dwt

Prop. (%)

fixed N to

zooplankton

(FtotalCy x

100)

15 % wwt to

dwt

Adjusted

transfer

10 % wwt to

dwt

Adjusted

transfer

2000 31 B1 0.563 31 % 40 % 18 % 23 %

2000 27 H4 0.762 69 % 76 % 53 % 59 %

Hypothesis E (Manuscript IV):

“Baltic Sea spring phytoplankton primary production is nitrogen limited.”

Here we present a 35 year long dataset (1977-2012), showing a clear shift in

temperature from colder (1977-1988) to warmer (1988-2012) temperatures

(Table 2). Also, at the coastal reference station (B1), DIN and DIP concentra-

tions displayed a clear breakpoint with increasing concentrations until 1993.

After 1993 until ~2006 DIN concentrations decreased, while DIP remained at

36

a high level (MSIV: Figs. 6 & 7). Phytoplankton spring primary productivity

varied between 24-60 gC m-2 in the coastal area (B1) and 29-57 gC m-2 for

the open sea (BY31). Average production at B1 in 1991-2012 (no reliable data

before 1991 at BY31) was 36.9 compared to 42.8 gC m-2, an estimated 16 %

higher production in the open sea.

Table 2: Mean monthly air temperature (oC) in the ‘cold’ and ‘warm’ periods

Year/month 1977-1987 1988-2012 Change

January -2.8 0.3 3.1

February -3.7 -0.3 3.4

March -0.7 1.2 1.9

April 2.4 3.9 1.5

We found a clear relation between spring 14C-uptake (a proxy for primary

productivity) and DIN depletion in spring particularly at B1 (Fig. 7, Table 4). At

BY31, the collinearity between DIN and DIP depletion made this relation less

clear, however BY31 was only sampled from 1991, after which no large

changes were observed for DIN and DIP depletion rates. Moreover, the shal-

low mean depth around the coastal reference station (B1) causes the mixed

layer frequently to be in contact with the sediments, influencing nutrient con-

centrations. Also, influences from both land and open may at this station mask

any collinearity between DIN and DIP uptake. However, the open sea phyto-

plankton C:P ratio was 17, which is considerably lower than the Redfield ratio

of 41 (w w-1), and the observed C:N ratios were 7.6 (B1) and 5.6 (BY31), which

are very close to the Redfield C:N ratio of 5.7. In addition, the particulate mat-

ter C:N and C:P ratios were 7 and 50 respectively (U. Larsson unpubl.). This

strongly indicates nitrogen limitation and not phosphorus limitation.

37

Figure 7: 14C-uptake vs. DIN depletion at the coastal (B1) and open sea (BY31) reference sta-

tions.

Table 4: Multiple linear regressions of spring primary productivity (PP), DIN and DIP uptake, water

temperature and salinity (mean Jan-April). Resp. var. is the response variable, most pars. model

is the most parsimonious model. Due to collinearity between DIN and DIP uptake at BY31, we

run two separate models with either DIN or DIP uptake included.

Resp. var.

Sta-tion

Start model

Most pars. model α t F

Adj. r2

AICstart AICend p-value

14C up-take

B1

water temp

- - - F1,31

= 14.6

0.30 226.4 224.4

-

DINuptake DINuptake 7.51 3.82 <0.001

DIPuptake - - - -

salinity - - - -

14C up-take

BY31

water temp

- - - F1,17

= 4.78

0.17 137.5 136.5

-

DINuptake DINuptake 5.86 2.19 >0.05

salinity - - - -

14C up-take

BY31

water temp

- - - F1,17

= 5.01

0.18 137.5 136.3

-

DIPuptake DIPuptake 20.4 2.24 >0.05

salinity - - - -

Hypothesis F (Manuscript IV):

“Diazotrophic nitrogen boosts summer phytoplankton primary production.

We found a significant positive effect of estimated mean summer N-fixation

(calculated from data in Ploug et al. (2011; 2010)) on mean summer primary

production at the open sea reference station (BY31), but not at the coastal

station (B1) (MSIV: Table 6). The lack of relation between these variables at

the coast is most likely due to the land influence as well as contact with the

sediments. The mean depth in this coastal area is merely 17 m resulting in a

high degree of remineralized DIN entering the mixed layer. This likely results

in a higher proportion diazotrophic-derived N of the total N at the open sea is

compared with the coast.

38

Hypothesis G (Manuscript IV):

“Filamentous diazotrophs have increased over the past decades in both

coastal and open sea areas due to changed temperatures and nutrient

concentrations”

We found that Aphanizomenon sp. has indeed increased at the coastal station

(linear regression: r2=0.66, p<0.0001), but not in the open sea area, where N.

spumigena and Dolichospermum spp. had increased significantly (linear re-

gression: r2=0.38, p<0.01 and r2=0.40, p<0.01, respectively), though the

patchy distribution of these species calls for caution in interpreting these re-

sults. Nonetheless, when when truncating the B1 dataset to include only the

years that were sampled at BY31, differences in patterns between the two

areas remained indicating fundamental differences at the coast compared with

the open sea. The increase in Aphanizomenon sp. in the coastal area may be

due to increased weather induced upwellings. The estimated summer nitrogen

fixation explains almost 40 % of the variability in total phytoplankton biomass

and over 30 % of variability in 14C-uptake during summer (Fig. 9). Due to four

outliers, there was no significant correlation between TN (see Methods) and 14C-uptake, but this method is very sensitive to water exchange.

39

Figure 9: The relation between two proxies for nitrogen fixation, TN (the increase in TN during

the cyanobacterial bloom) and N-fixation estimated from Aphanizomenon sp. and Nodularia spu-

migena biomass, as well as total phytoplankton biomass in addition to measured 14C-uptake.

I sincerely hope that our findings may be of help in supporting good manage-

ment decisions and be part of a knowledge base for political decisions on

sewage treatment and land use in the EU in the future.

40

Acknowledgements and Funding

First and foremost, many thanks to my main supervisor Ulf Larsson for discus-

sions, ideas and support. I also would like to thank Jakob Walve, my first co-

supervisor for invaluable help in the field. Also, many thanks to Elena

Gorokhova, my second co-supervisor for advice and a new look on things. I am

grateful for the microscopy help received from Susanna Hajdu and Helena

Höglander. In addition, I would like to express my gratitude to Helena Höglan-

der for sampling for me the summer of 2008 when I was on parental leave.

The chemistry laboratory, at the former Department of Systems Ecology,

has conducted some splendid work included in this thesis such as nutrient anal-

yses and some parts of the colony nutrient analysis. You are indeed of one of

the best laboratories in Europe. Without you, this thesis would not exist. The

crews of Limanda and Aurelia have shipped me a fair few nautical miles these

years. Thank you! Thank you to Pelle and Hedvig for the summer of 2007. I

also would like to thank the administrative personnel, the personnel at SYVAB

and the Stable Isotope Laboratory (SIL) at Stockholm University for help and

good instructions.

Many thanks to Antonia Sandman and Martin Ogonowski for statistical and

modeling discussions. Antonia: you are a great friend! Also, thank you Sture

Hansson for being a friend and for provoking new ideas. Many thanks to eve-

ryone at the previous Department of Systems Ecology (now Department of

Ecology, Environment and Plant Sciences).

Thanks to my favorite wine- and coffee producers, without you, I would have

been lost. I am also grateful to Susanne Diemer and her team at Yggdra-

silDiemer, Berlin, Germany for office space, professional discussions and

friendship. An extra thank you to Ragnar Elmgren for invaluable comments and

for taking time to read, even though time was short.

I would also like to express my gratitude to the two internal reviewers: Ulla

Rasmusson and Monika Winder for taking time to read this thesis. Also, to Mi-

caela Hellström, Carina O’Reilly, Joanne Kyle and Antonia Sandman for read-

ing and commenting and for being fantastic friends. Thank you!

The project was financed by Sydvästra Stockholmsregionens Vatten-

verkaktiebolag (SYVAB) (Himmerfjärden Eutrophication Study) by grants to Dr.

41

Ulf Larsson and by FORMAS for the project “Managing Baltic nutrients in rela-

tion to cyanobacterial blooms: What should we aim for?” to Dr. Ragnar

Elmgren.

42

References

Adams, D. G. 2000. Heterocyst formation in cyanobacteria. Current Opinion in Mi-

crobiology 3: 618-624.

Adams, D. G., and P. S. Duggan. 1999. Tansley Review no. 107. Heterocyst and

akinete differentiation in cyanobacteria. New Phytologist 144: 3–33.

Almesjo, L., and C. Rolff. 2007. Automated measurements of filamentous cyanobac-

teria by digital image analysis. Limnology and Oceanography-Methods 5:

217-224.

Altabet, M. A. and others 1999. The nitrogen isotope biogeochemistry of sinking par-

ticles from the margin of the Eastern North Pacific. Deep-Sea Research Part

I-Oceanographic Research Papers 46: 655-679.

Bianchi, T. S., B. Johansson, and R. Elmgren. 2000. Breakdown of phytoplankton

pigments in Baltic sediments: effects of anoxia and loss of deposit-feeding

macrofauna. Journal of Experimental Marine Biology and Ecology 251: 161-

183.

Bothe, H. 1982. Nitrogen fixation. In The Biology of Cyanobacteria,.

Bottomley, P. J., J. F. Grillo, C. Van Baalen, and F. R. Tabita. 1979. Synthesis of

nitrogenase and heterocysts by Anabaena sp. CA in the presence of high lev-

els of ammonia. Journal of Bacteriology 140: 938–943.

Brett, M. T., and D. C. Muller-Navarra. 1997. The role of highly unsaturated fatty

acids in aquatic food web processes. Freshwater Biology 38: 483-499.

Brettar, I., and G. Rheinheimer. 1991. Denitrification in the Central Baltic - Evidence

for H2S-Oxidation as Motor of Denitrification at the Oxic-Anoxic Interface.

Marine Ecology-Progress Series 77: 157-169.

Burnham, K. P., and D. R. Anderson. 2004. Multimodel inference - understanding

AIC and BIC in model selection. Sociological Methods & Research 33: 261-

304.

Carmichael, W. W. 1992. Cyanobacteria secondary metabolites - the cyanotoxins.

Journal of Applied Bacteriology 72: 445-459.

Conley, D. J. and others 2009. Tackling Hypoxia in the Baltic Sea: Is Engineering a

Solution? Environmental Science & Technology 43: 3407-3411.

Dumont, H., I. Van De Velde, and S. Dumont. 1975. The dry weight estimate of bio-

mass in a selection of Cladocera, Copepoda and Rotifera from the plankton,

periphyton and benthos of continental waters. Oecologia 19: 75-97.

Elmgren, R., and C. Hill. 1997. Ecosystem function at low biodiversity: the Baltic

example. Cambridge University Press.

Engqvist, A. 1996. Long-term nutrient balances in the eutrophication of the Him-

merfjarden estuary. Estuarine Coastal and Shelf Science 42: 483-507.

Flores, E., and A. Herrero. 1994. Assimilatory nitrogen metabolism and its regulation.

43

Flores, E., A. Herrero, C. P. Wolk, and I. Maldener. 2006. Is the periplasm continuous

in filamentous multicellular cyanobacteria? Trends in Microbiology 14: 439-

443.

Fry, B. 2006. Stable isotope ecology. p. 140. Springer science+business media.

Gorokhova, E., and J. Engstrom-Ost. 2009. Toxin concentration in Nodularia spumi-

gena is modulated by mesozooplankton grazers. Journal of Plankton Re-

search 31: 1235-1247.

Granéli, E., K. Wallström, U. Larsson, W. Granéli, and R. Elmgren. 1990. Nutrient

limitation of primary production in the Baltic Sea area. Ambio 19: 142-151.

Guerrero, M. G., and C. Lara. 1987. The Cyanobacteria: Assimilation of inorganic

nitrogen. Amsterdam. Elsevier Science. Fay, P., Van Baalen, C. pp. 163–

185.

Gustafsson, B. G. and others 2012. Reconstructing the Development of Baltic Sea

Eutrophication 1850-2006. Ambio 41: 534-548.

Hajdu, S., H. Höglander, and U. Larsson. 2007. Phytoplankton vertical distributions

and composition in Baltic Sea cyanobacterial blooms. Harmful Algae 6: 189-

205.

Hansson, M., and L. Andersson. 2013. Oxygen Survey in the Baltic Sea 2013 -Extent

of anoxia and hypoxia, 1960 - 2013, SMHI Report No. 49, 2013. In HEL-

COM [ed.], HELCOM Baltic Sea Environment Fact Sheets. HELCOM.

Heiskanen, A. S., and K. Kononen. 1994. Sedimentation of vernal and late summer

phytoplankton communities in the coastal Baltic Sea. Arch. Hydrobiologia:

175-198.

Helcom. 1988. Biovolumes and Size-Classes of Phytoplankton in the Baltic Sea. Bal-

tic Sea Environment Proceedings No.106. Helsinki Commission Baltic Ma-

rine Environment Protection Commission.

Hense, I., and A. Beckmann. 2006. Towards a model of cyanobacteria life cycle -

effects of growing and resting stages on bloom formation of N2-fixing spe-

cies. Ecological Modelling 195: 205-218.

Hernroth, L. 1985. Recommendations on methods for marine biological studies in the

Baltic Sea. Mesozooplankton assessment. BMB Publication 10: 1-32.

Hogfors, H. and others 2014. Bloom-Forming Cyanobacteria Support Copepod Re-

production and Development in the Baltic Sea. PLoS One 9.

Holl, C. M., and J. P. Montoya. 2005. Interactions between nitrate uptake and nitrogen

fixation in continuous cultures of the marine diazotroph Trichodesmium (Cy-

anobacteria). Journal of Phycology 41: 1178-1183.

Hothorn, T., K. Hornik, M. A. Van De Wiel, and A. Zeileis. 2008. Implementing a

Class of Permutation Tests: The coin Package. . Statistical Software 28: 1-

23.

Howarth, R., and H. W. Paerl. 2008. Coastal marine eutrophication: Control of both

nitrogen and phosphorus is necessary. Proceedings of the National Academy

of Sciences of the United States of America 105: E103-E103.

Huber, A. L. 1984. Nodularia (cyanobacteriaceae) akinetes in the sediments of the

Peel-Harvey Estuary, Western Australia:potential inoculum source for Nod-

ularia blooms. Applied and Environmental Microbiology 47: 234-238.

---. 1985. Factors affecting the germination of akinetes of Nodularia spumigena (Cy-

anobacteriaceae). Applied and Environmental Microbiology 49: 73-78.

44

Höglander, H., U. Larsson, and S. Hajdu. 2004. Vertical distribution and settling of

spring phytoplankton in the offshore NW Baltic Sea proper. Marine Ecology-

Progress Series 283: 15-27.

Kahru, M., and R. Elmgren. 2014. Multidecadal time series of satellite-detected accu-

mulations of cyanobacteria in the Baltic Sea. Biogeosciences 11: 3619-3633.

Karhu, M., U. Horstmann, and O. Rud. 1994. Satellite detection of increased cyano-

bacterial blooms in the Baltic Sea: Natural fluctuations or ecosyste change?

Ambio 23: 469-472.

Karjalainen, M. and others 2007. Ecosystem consequences of cyanobacteria in the

northern Baltic Sea. Ambio 36: 195-202.

Karjalainen, M., M. Reinikainen, F. Lindvall, L. Spoof, and J. A. O. Meriluoto. 2003.

Uptake and accumulation of dissolved, radiolabeled nodularin in Baltic Sea

Zooplankton. Environmental Toxicology 18: 52-60.

Koroleff, F. 1983. Determination of nutrients. In: Grasshoff K, Ehrhardt M, Kremling

K (eds) Methods of seawater analysis. Verlag Chemie, Weinheim, p. 125–

139 (phosphorus method), p. 162–173 (nitrogen method). Kremling K, Pe-

tersen H (1984) Synoptic survey.

Kozlowsky-Suzuki, B. and others 2007. Disruption of the microbial food web and

inhibition of metazooplankton development in the presence of iron- and

DOM-stimulated Baltic Sea cyanobacteria. Marine Ecology-Progress Series

337: 15-26.

Krauss, W., and B. Brügge. 1991. Wind-Produced Water Exchange between the Deep

Basins of the Baltic Sea. Journal of Physical Oceanography 21: 373-384.

Laamanen, M., H. Kuosa, and S. Maximum. 2005. Annual variability of biomass and

heterocysts of the N-2-fixing cyanobacterium Aphanizomenon flos-aquae in

the Baltic Sea with reference to Anabaena spp. and Nodularia spumigena.

Boreal Environment Research 10: 19-30.

Lampert, W. 1987. Laboratory studies on zooplankton-cyanobacteria interactions. N.

Z. J. Mar. Freshwat. Res. 21: 483– 490.

Larsson, U., R. Elmgren, and F. Wulff. 1985. Eutrophication and the Baltic Sea:

causes and consequences. Ambio 14: 10-14.

Larsson, U., S. Hajdu, J. Walve, and R. Elmgren. 2001. Baltic Sea nitrogen fixation

estimated from the summer increase in upper mixed layer total nitrogen.

Limnology and Oceanography 46: 811-820.

Lehtimäki, J., P. Moisander, K. Sivonen, and K. Kononen. 1997. Growth, nitrogen

fixation, and nodularin production by two Baltic sea cyanobacteria. Applied

and Environmental Microbiology 63: 1647-1656.

Lindahl, G., K. Wallstrom, and G. Brattberg. 1980. Short-term variations in nitrogen-

fixation in a coastal area of the northern Baltic. Archiv Für Hydrobiologie

89: 88-100.

Macko, S. A., M. L. Fogel, P. E. Hare, and T. C. Hoering. 1987. Isotopic fractionation

of nitrogen and carbon in the synthesis of amino-acids by microorganisms.

Chemical Geology 65: 79-92.

Mayhew, S. 2004. A Dictionary of Geography.

Mendoza, J. J. 1993. A preliminary budget for the North Eastern Venezuela Shelf

Ecosystem International Center for Living Aquatic Resources Management.

Minagawa, M., and E. Wada. 1984. Stepwise enrichment of N-15 along food-chains

- further evidence and the relation between Delta N-15 and animal age. Ge-

ochimica Et Cosmochimica Acta 48: 1135-1140.

45

Mulholland, M. R., D. A. Bronk, and D. G. Capone. 2004. Dinitrogen fixation and

release of ammonium and dissolved organic nitrogen by Trichodesmium

IMS101. Aquatic Microbial Ecology 37: 85-94.

Ohlendieck, U., K. Gundersen, M. Meyerhofer, P. Fritsche, K. Nachtigall, and B.

Bergmann. 2007. The significance of nitrogen fixation to new production

during early summer in the Baltic Sea. Biogeosciences 4: 63-73.

Ohlendieck, U., A. Stuhr, and H. Siegmund. 2000. Nitrogen fixation by diazotrophic

cyanobacteria in the Baltic Sea and transfer of the newly fixed nitrogen to

picoplankton organisms. Journal of Marine Systems 25: 213-219.

Olenina, I. and others 2006. Biovolumes and size-classes of phytoplankton in the Bal-

tic Sea, p. 144. HELCOM Baltic Sea Environmental Proceedings.

Ploug, H. and others 2011. Carbon, nitrogen and O-2 fluxes associated with the cya-

nobacterium Nodularia spumigena in the Baltic Sea. Isme Journal 5: 1549-

1558.

---. 2010. Carbon and nitrogen fluxes associated with the cyanobacterium Aphani-

zomenon sp. in the Baltic Sea. Isme Journal 4: 1215-1223.

R CoreTeam, 2008. R: A language and environment for statistical computing. R Foun-

dation for Statistical Computing.

Rydén, L., P. Migula, and M. Andersson. 2003. Environmental science. Baltic Uni-

versity Press.

Sanz-Alférez, S., and F. F. Del Campo. 1994. Relationship between nitrogen fixation

and nitrate metabolism in the NodMaria strains Nil and M2. Planta: 339-345.

Sarkar, D. 2008. lattice. Springer.

Sarkar, D., and F. Andrews. 2013. latticeExtra.

Schindler, D. W., and R. E. Hecky. 2008. Reply to Howarth and Paerl: Is control of

both nitrogen and phosphorus necessary? Proceedings of the National Acad-

emy of Sciences of the United States of America 105: E104-E104.

Schindler, D. W. and others 2008. Eutrophication of lakes cannot be controlled by

reducing nitrogen input: Results of a 37-year whole-ecosystem experiment.

Proceedings of the National Academy of Sciences of the United States of

America 105: 11254-11258.

Schinke, H., and W. Matthaus. 1998. On the causes of major Baltic inflows - an anal-

ysis of long time series. Continental Shelf Research 18: 67-97.

Schulz, S., and W. Kaiser. 1986. Increasing trends in plankton variables in the Baltic

Sea - a further sign of eutrophication? Ophelia: 249-257.

Sellner, K. G., G. J. Doucette, and G. J. Kirkpatrick. 2003. Harmful algal blooms:

causes, impacts and detection. J Ind Microbiol Biotechnol 30: 383-406.

Smayda, T. J. 1997. What is a bloom? A commentary. Limnology and Oceanography

42: 1132-1136.

Sohm, J. A., E. A. Webb, and D. G. Capone. 2011. Emerging patterns of marine ni-

trogen fixation. Nature Reviews Microbiology 9: 499-508.

Sommer, U. 1988. Does nutrient competition among phytoplankton occur in situ?

Verh. internat. Limnol.: 707-712.

Stal, L. J. and others 2003. BASIC: Baltic Sea cyanobacteria. An investigation of the

structure and dynamics of water blooms of cyanobacteria in the Baltic Sea-

responses to a changing environment. Continental Shelf Research.: 1695-

1714.

Stal, L. J., M. Staal, and M. Villbrandt. 1999. Nutrient control of cyanobacterial

blooms in the Baltic Sea. Aquatic Microbial Ecology 18: 165-173.

46

Stal, L. J., and A. E. Walsby. 2000. Photosynthesis and nitrogen fixation in a cyano-

bacterial bloom in the Baltic Sea. European Journal of Phycology 35: 97-

108.

Stephens, N., K. J. Flynn, and J. R. Gallon. 2003. Interrelationships between the path-

ways of inorganic nitrogen assimilation in the cyanobacterium Gloeothece

can be described using a mechanistic mathematical model. New Phytologist

160: 545-555.

Ståhl-Deblanco, A., L.-A. Hansson, and M. Gyllström. 2003. Recruitment of resting

stages may induce blooms of Microcystis at low N:P ratios. JOURNAL OF

PLANKTON RESEARCH 25: 1099-1106.

Suikkanen, S., H. Kaartokallio, S. Hallfors, M. Huttunen, and M. Laamanen. 2010.

Life cycle strategies of bloom-forming, filamentous cyanobacteria in the

Baltic Sea. Deep-Sea Research Part Ii-Topical Studies in Oceanography 57:

199-209.

Taiz, L., and E. Zeiger. 2002. Plant Physiology, 3 ed.

Thompson, P.-A., H.-M. Oh, and G.-Y. Rhee. 1994. Storage of phosphorus in nitrogen

fixing Anabaena flos aquae (Cyanophyceae). Journal of Phycology: 267-

273.

Utne-Palm, A. C. 2002. Visual feeding of fish in a turbid environment: Physical and

behavioural aspects. Marine and Freshwater Behaviour and Physiology 35:

111-128.

Wacklin, P., L. Hoffmann, and J. Komarek. 2009. Nomenclatural validation of the

genetically revised cyanobacterial genus Dolichospermum (RALFS ex

BORNET et FLAHAULT) comb. nova. Fottea 9: 59-64.

Walsby, A. E., P. K. Hayes, and R. Boje. 1995. The gas vesicles, buoyancy and ver-

tical distribution of cyanobacteria in the Baltic Sea. European Journal of Phy-

cology 30: 87 — 94.

Walve, J. 2002. Nutrient limitation and elemental ratios in Baltic Sea plankton. Stock-

holm University.

Walve, J., and U. Larsson. 2007. Blooms of Baltic Sea Aphanizomenon sp (cyano-

bacteria) collapse after internal phosphorus depletion. Aquatic Microbial

Ecology 49: 57-69.

Wasmund, N. 1997. Occurrence of cyanobacterial blooms in the Baltic Sea in relation

to environmental conditions. Internationale Revue gesamten Hydrobiologie

82: 169-184.

Wasmund, N., M. Voss, and K. Lochte. 2001. Evidence of nitrogen fixation by non-

heterocystous cyanobacteria in the Baltic Sea and re-calculation of a budget

of nitrogen fixation. Marine Ecology-Progress Series 214: 1-14.

Wickham, H. 2009. ggplot2: elegant graphics for data analysis. Springer New York.

---. 2011. The Split-Apply-Combine Strategy for Data Analysis. Journal of Statistical

Software 40: 1-29.

Vintila, S., and R. El-Shehawy. 2007. Ammonium ions inhibit nitrogen fixation but

do not affect heterocyst frequency in the bloom-forming cyanobacterium

Nodularia spumigena strain AV1. Microbiology–Society for General Micro-

biology 153: 3704-3712.

Wood, S. N. 2003. Thin-plate regression splines. Journal of the Royal Statistical So-

ciety (B) 65: 95-114.

47

Xu, Z. K., and C. W. Burns. 1991. Development, growth and survivorship of juvenile

calanoid copepods on diets of cyanibacteria and algae. Internationale Revue

Der Gesamten Hydrobiologie 76: 73-87.

Yamamoto, Y. 1995. Effect of desiccation on the germination of akinetes of Anabaena

cylindrical. Plant Cell Physiology 16: 749-752.

Zakrisson, A., and U. Larsson. 2014. Regulation of heterocyst frequency in Baltic Sea

Aphanizomenon sp. Journal of Plankton Research.

Zakrisson, A., U. Larsson, and H. Höglander. 2014. Do Baltic Sea diazotrophic cya-

nobacteria take up combined nitrogen in situ? Journal of Plankton Research.

Zeileis, A., and G. Grothendieck. 2005. zoo: S3 Infrastructure for Regular and Irreg-

ular Time Series. Journal of Statistical Software 14: 1-27.

Zuur, A. F., E. N. Ieno, N. J. Walker, A. A. Saveliev, and G. M. Smith. 2009. Mixed

Effects Models and Extensions in Ecology with R. Springer Verlag.