Thesis
Continental scale cyanobacterial dynamics under global warming and eutrophication
MANTZOUKI, Evanthia
Abstract
On-going global warming and eutrophication are expected to promote cyanobacterial dominance worldwide. Although increased lake temperature and nutrients are well-established drivers of blooms, the mechanisms that determine cyanobacterial biomass are complex, with potentially direct, indirect and interactive effects. Cyanobacteria can produce toxins that constitute a considerable risk for animal and human health. Such global range phenomena should be studied at a wide spatial scale, to directly compare phytoplankton response in different lake types across contrasting climatic zones. During this dissertation, the European Multi-lake survey (EMLS) was organised in order to sample lakes across Europe and disentangle the effect of environmental stressors on potentially toxic cyanobacterial blooms. The results demonstrated that the distribution of cyanobacterial toxins and the toxic potential in lakes was highly dependent on direct and indirect effects of temperature. Nutrients interacted synergistically with increased lake temperatures to promote cyanobacterial growth more than that of other phytoplankton taxa. Providing [...]
Reference
MANTZOUKI, Evanthia. Continental scale cyanobacterial dynamics under global warming and eutrophication. Thèse de doctorat : Univ. Genève, 2018, no. Sc. 5275
DOI : 10.13097/archive-ouverte/unige:112209 URN : urn:nbn:ch:unige-1122091
Available at: http://archive-ouverte.unige.ch/unige:112209
Disclaimer: layout of this document may differ from the published version.
1 / 1 UNIVERSITÉ DE GENÈVE FACULTÉ DES SCIENCES Section des Sciences de la Terre et de l’Environnement Département F.-A. Forel des Sciences Professeur Dr. Bastiaan W. Ibelings de l’Environnement et de l’Eau Institut des Sciences de l’Environnement
Continental Scale Cyanobacterial Dynamics under Global Warming and Eutrophication
THÈSE
présentée à la Faculté des Sciences de l’Université de Genève pour obtenir le grade de Docteur ès sciences, mention sciences de l’environnement
par Evanthia MANTZOUKI
de Thessalonique (Grèce)
Thèse N° 5275
GENÈVE Atelier d’impression ReproMail l’ Université de Genève 2018
1
Table of Contents RÉSUMÉ ...... 5 CHAPTER 1 ...... 8 The principle and value of the European Multi Lake Survey ...... 8 Abstract ...... 9 Introduction ...... 9 Technological progress ...... 11 Scientific Responsibility ...... 11 Research Questions ...... 13 Acknowledgements ...... 16 CHAPTER 2 ...... 17 Understanding the key ecological traits of cyanobacteria as a basis for their management and control in changing lakes ...... 17 Abstract ...... 18 Introduction ...... 19 Cyanobacteria and their key traits ...... 23 Functional classification of cyanobacteria ...... 25 Conclusions and research perspectives ...... 35 Acknowledgments ...... 36 CHAPTER 3 ...... 37 A European Multi Lake Survey dataset of environmental variables, phytoplankton pigments and cyanotoxins ...... 37 Abstract ...... 38 Background and summary ...... 38 Methods ...... 40 Data records ...... 48 Technical validation ...... 49 Data citations ...... 50 Acknowledgements ...... 52 Author contributions ...... 52 CHAPTER 4 ...... 53
2 A European Multi Lake Survey in one of the hottest summers on record: cooler Boreal regions develop bigger blooms than warmer Mediterranean regions ...... 53 Abstract ...... 54 Introduction ...... 55 Materials & Methods ...... 57 Results ...... 61 Discussion ...... 70 Acknowledgements ...... 78 Supporting material ...... 78 CHAPTER 5 ...... 84 Temperature effects explain continental scale distribution of cyanobacterial toxins ...... 84 Abstract ...... 85 Introduction ...... 86 Results ...... 88 Discussion ...... 96 Materials & Methods ...... 101 Supplementary Material ...... 107 Acknowledgements ...... 107 Author Contributions ...... 107 CHAPTER 6 ...... 108 Opinion: Multi-lake snapshot surveys for lake monitoring, more than a shot in the dark ...... 108 Abstract ...... 109 Introduction - why do we monitor? ...... 110 Different monitoring strategies ...... 110 Advantages of the MLSS ...... 112 Conclusions ...... 115 Abbreviations ...... 116 Acknowledgements ...... 116 Author Contributions ...... 116 CHAPTER 7 ...... 117 Conclusions and Perspectives ...... 117 References ...... 124
3 Remerciements ...... 161 Acknowledgements (English version) ...... 164 Authors and Affiliations ...... 167 CV & List of Publications ...... 176 EMLS Dataset ...... 178
4 RÉSUMÉ
Dynamiques cyanobactériennes à l’échelle continentale sous l’eutrophisation et le changement climatique L'eutrophisation reste un problème important pour les systèmes aquatiques en Europe. Les proliférations de cyanobactéries étouffent d'autres formes de vie aquatique et produisent des toxines qui mettent en danger la production d'eau potable, de loisirs et de santé publique. La littérature scientifique récente prévoit que le réchauffement climatique augmentera ces problèmes. Un climat plus chaud favorisera les efflorescences cyanobactériennes et menace de défaire des décennies de restauration réussie des lacs en Europe, où des mesures visant à atténuer l'eutrophisation et à contrôler les efflorescences cyanobactériennes ont été prises au détriment de milliards d'euros. Pourtant, il n'y a pas de consensus scientifique sur la façon dont l'eutrophisation et le changement climatique vont interagir. Certains auteurs affirment que les nutriments resteront la cause principale des efflorescences et ne verront qu'un rôle mineur pour la température, alors que d'autres prédisent que le réchauffement climatique aura un impact majeur (« blooms-like-it-hot »). La question la plus importante: les nutriments et la température interagiront-ils en synergie ? Ceci n'est pas seulement d'un intérêt scientifique, ça détermine également les mesures qui doivent être prises pour préparer nos écosystèmes et contrôler les proliférations à l'avenir. Qu'est-ce qui facilite la prolifération des cyanobactéries? Les cyanobactéries ont de nombreux traits écologiques (fonctionnels) qui peuvent leur permettre de prospérer dans divers scénarios de changement environnemental. Ces caractéristiques peuvent varier de façon prévisible selon les gradients environnementaux, comme la régulation de la flottabilité pendant les périodes de stratification accrue ou la fixation de l'azote lors de la limitation de l'azote. Un autre trait que possèdent de nombreux taxons cyanobactériens est la production de divers types de toxines, telles que les hépatotoxines, les neurotoxines et les cytotoxines. Des expériences ont montré que l'augmentation de la température de l'eau et des nutriments peut interagir en synergie pour stimuler la production de souches toxiques (Lürling et al. 2017). Au chapitre 2, nous utilisons la classification fonctionnelle et la littérature scientifique de Reynolds (2006) pour rassembler des informations sur la manière dont certains groupes de cyanobactéries répondent principalement aux perturbations environnementales. Nous nous axons sur cinq groupes de cyanobactéries qui incluent des espèces qui sont généralement notoires pour la formation de fleurs d’eau. Nous présentons leurs principales caractéristiques et comment ces caractéristiques maintiendront leur fonction dans le cadre de futurs changements environnementaux. Comprendre comment et quand ces traits fonctionnels entrent en jeu et faciliter la domination des espèces cyanobactériennes pourrait révéler le "talon d’Achille" de ce groupe
5 fonctionnel, aider les gestionnaires des lacs à perturber les conditions qui favorisent leur croissance exponentielle et éventuellement les contrôler (Mantzouki et al. 2016). Étant donné que les proliférations de cyanobactéries sont une réponse typique des systèmes aquatiques à la perturbation de l’environnement dans le monde, il est urgent d’établir les schémas d’apparition des cyanobactéries à l’échelle régionale, continentale et mondiale. Comprendre l'impact des phénomènes à l'échelle mondiale nécessite des informations provenant de nombreux lacs présentant des caractéristiques similaires (par exemple, morphométrie, statut trophique) sur une large échelle géographique pour démontrer s'ils répondent de manière cohérente à une pression environnemental similaire. Cette thèse doctorale a organisé une enquête sur plusieurs lacs en Europe (European Multi Lake Survey-EMLS) au cours de l'été 2015. Pendant l’enquête d’EMLS, 27 pays européens ont échantillonné environ 400 lacs pour des obtenir des données chimiques, physiques et biologiques d'une manière absolument standardisée afin d'assurer la comparabilité et l'intégration des données (de plus amples informations sur l’organisation d’EMLS figurent au Chapitre 3). L’EMLS a mis l’accent sur l’importance des effets directs et indirects du réchauffement planétaire et de l’eutrophisation, ainsi que sur le mode d’interaction entre eux (Chapitre 4). Bien que les expériences et les études de modélisation suggèrent clairement une interaction synergique possible entre l'augmentation des nutriments et la température dans la prolifération des cyanobactéries (Rigosi et al. 2014) et leur production de toxines (Lürling et al. 2017), les preuves provenant des observations sur le terrain sont à peine présentes. La stabilité de la colonne d'eau (stratification) devrait également augmenter en tant qu'effet indirect de l'augmentation de la température (Gerten et Adrian, 2002). Une stratification améliorée peut établir une couche mélangée peu profonde proche de la surface, limitant la disponibilité de la lumière dans cette zone étroite. La disponibilité de la lumière ne favorisera que le phytoplancton capable de maintenir sa position - comme les espèces cyanobactériennes flottantes (Reynolds, 2006) - dans cette zone étroite. Ainsi, le climat légèrement modifié (effet indirect des nutriments élevés) pourrait également interagir en synergie avec une stratification accrue, accentuant les proliférations cyanobactériennes à l’échelle continentale. L'EMLS visait à répondre à cette question en échantillonnant principalement des lacs (hyper) eutrophiques. L'analyse de l’interaction entre les nutriments et la température peut être utile pour déterminer si et comment (de manière synergique, additive ou antagoniste) varie la biomasse totale des algues, cyanobactéries et algues vertes sur le continent. Nous avons également testé si l'effet des variables environnementales susmentionnées sur la biomasse algale ou cyanobactérienne différait entre lacs de différents types de profondeur (peu profond par rapport à profond) ou emplacement géographique basé sur différentes zones climatiques (Méditerranée versus Continentale versus Boréale). Un aperçu de la manière dont les changements environnementaux déterminent la production et la distribution des toxines cyanobactériennes (hépatotoxines, neurotoxines, cytotoxines) est nécessaire pour l'évaluation et la gestion des risques
6 (Ibelings et al. 2014). Néanmoins, les études expérimentales ne sont pas encore concluantes pour attribuer la production de toxines à des espèces cyanobactériennes spécifiques dans des conditions environnementales spécifiques (Neilan et al. 2013). Par conséquent, le manque de cohérence des résultats expérimentaux et de la normalisation des campagnes de surveillance et d'échantillonnage empêche de comprendre comment les facteurs de stress environnementaux sont liés à la production de toxines cyanobactériennes. Au Chapitre 5, nous utilisons des approches d'écologie communautaire pour examiner les toxines comme une « communauté» de toxines potentiellement coexistantes, au lieu de chaque toxine séparément. Les effets directs et indirects de la température et des nutriments sont testés pour déterminer leur importance dans l'explication de la distribution des différentes toxines à l'échelle du continent. Un indice de diversité des toxines (TDI) est calculé en utilisant l'équation de Shannon et testé sur toutes les variables environnementales pour déterminer comment les mélanges de toxines se formeront sur le continent et si la perturbation environnementale pourrait favoriser quelques souches hautement toxiques. Une plus grande diversité de toxines pourrait peut-être résulter à une plus grande stabilité de la toxicité globale dans une prolifération, car si une toxine décline, une autre peut augmenter, entraînant une persistance de la toxicité globale. Alors que le réchauffement climatique perpétue et que l'eutrophisation persiste, les effets directs et indirects de l'augmentation des températures et des nutriments dans les lacs entraîneront des changements dans la biomasse cyanobactérienne et la distribution des cyanotoxines en Europe.
7 CHAPTER 1 The principle and value of the European Multi Lake Survey
Mantzouki and Ibelings, L & O Bulletin, 2018
8 1Abstract On-going global warming and eutrophication are expected to promote cyanobacterial dominance worldwide. Although increased lake temperature and nutrients are well- established drivers of blooms, the mechanisms that determine cyanobacterial biomass are complex, with potentially direct, indirect and interactive effects. Cyanobacteria can produce toxins that constitute a considerable risk for animal and human health and thus a substantial economic cost if we are to ensure safe drinking water. Such global range phenomena should be studied at a wide spatial scale, to directly compare phytoplankton response in different lake types across contrasting climatic zones. The European Multi Lake survey sought to harness the power of group science in order to sample lakes across Europe and disentangle the effect of environmental stressors on potentially toxic cyanobacterial blooms. The first EMLS results showed that the distribution of cyanobacterial toxins and the toxic potential in lakes will be highly dependent on direct and indirect effects of temperature. If nutrients are not regulated, then they may interact synergistically with increased lake temperatures to promote cyanobacterial growth more than that of other phytoplankton taxa. Providing continental scale evidence is highly significant for the development of robust models that could predict cyanobacterial or algal response to environmental change. Introduction Cyanobacteria, the blue-green superstars of the aquatic world, have been around for a long time; ever since they pioneered the use of sunlight to make food and produce oxygen as a fortunate waste product. This so-called “Great Oxygenation Event” or “Oxygen Catastrophe”, brought with it mass extinction, as excess oxygen was toxic for anaerobic bacteria (Young, 2012). Life eventually adapted, and aerobic bacteria started to appear, taking up oxygen from the atmosphere and playing their part in regulating oxygen to levels that we have today. The evolution of photosynthetic organisms enabled life to diversify, and led to the biological world as we know it from fossil records and modern-day observations. Nowadays, cyanobacteria are again the culprits of ecosystem imbalance, this time by promoting anoxia and biodiversity loss, supporting a notion of “Cyanobacteria giveth, and cyanobacteria taketh away”. Catchy scientific articles, such as “Blooms like it hot” (Paerl and Huisman, 2008), “Blooms bite the hand that feeds them” (Paerl and Otten, 2013), “Is the future blue- green?” (Elliott 2012), along with the controversial issue “It takes two to Tango” (Paerl et al. 2016) in relation to which nutrient – N + P or P-only? (also see Schindler et al. 2016) - to control, and the prominent “Allied attack” (Moss et al. 2011) of climate warming and eutrophication, spotlight their notoriety. Anticipated climatic changes
1 This chapter has been published in Limnology and Oceanography bulletin, as part of the Chapter 1 Challenge that was organised in ASLO Victoria-2018. Here, we incorporate several modifications throughout the paper in order to create a more natural flow in the thesis manuscript. The original article can be found online by following the link: https://doi.org/10.1002/lob.10259.
9 along with on-going eutrophication, are expected to promote cyanobacterial dominance worldwide either directly or indirectly. The direct effects of these stressors mainly address the physiological properties of organisms for example by facilitating growth through nutrient availability and warmer temperatures. The indirect effects of such stressors modify the physical environment, like enhancing the stability of the water column, which can support the growth of buoyant cyanobacterial species. Climate warming and eutrophication may also interact synergistically and thus intensify cyanobacterial blooms. Answers to what drives cyanobacterial blooms or cyanobacterial toxin production, and how can we deal with them, are rather complicated and typically system specific. Consensus among scientists may be hard to achieve since often there are valid arguments from either side, excluding a single solution to ecological problems. The scientific community needs to take a step forward in changing the way we assume responsibility towards nature. “It takes a village to raise a child”. Even if we cannot predict with point accuracy the status of the ecosystems in a few years, we do know for a fact that climate is changing, temperature is rising and environmental catastrophes are becoming more frequent. Previous studies have demonstrated that cyanobacteria have the capacity to outcompete other primary producers as they proliferate from environmental disturbances (review in Carey et al. 2012). Gathering more information at a global scale, to demonstrate how patterns are consistently repeated, can help reduce uncertainty and eventually push for stricter regulation to ensure freshwater quality. What makes cyanobacteria so capable? Cyanobacteria have many ecological (functional) traits that may allow them to thrive under various scenarios of environmental change. These traits can vary predictably along environmental gradients like buoyancy regulation during periods of enhanced stratification, or nitrogen fixation during nitrogen limitation. Another trait that many cyanobacterial taxa possess is the production of various types of toxins, such as hepatotoxins, neurotoxins and cytotoxins. These toxins constitute a considerable risk and a substantial economic cost in achieving safe drinking or recreational water (Codd et al. 2017). Experiments have shown that increased water temperatures and nutrients can interact synergistically to boost the production of toxic strains (Lürling et al. 2017). In Chapter 2, we use Reynolds’ (2006) functional classification and scientific literature to gather information on how certain cyanobacteria groups mostly respond to environmental perturbation. We focus on five groups of cyanobacteria that include species that are typically notorious for bloom formation. We present their key traits and how these characteristics will maintain their function under future environmental change. Understanding how and when those functional traits come into play and facilitate the dominance of cyanobacterial species could reveal the “Achilles heel” of the particular functional group, aid lake managers to disrupt the conditions that favour their exponential growth and eventually control them (Mantzouki et al. 2016). Since cyanobacterial blooms are a typical response of aquatic systems to
10 environmental perturbation worldwide, it is urgent to draw the patterns of cyanobacterial occurrence at regional, continental and global scale. Technological progress Nowadays, as analytical methods become more sensitive they produce much finer results that can lead to better ecological understanding. Although such advances can increase the information output, they can also be overwhelming and either hamper or bias our ecological understanding. For example, increased biodiversity in a long-term phytoplankton dataset may have been the outcome of enhanced stratification and spatial niche diversification (Pomati et al. 2012) or increased experience in recognizing phytoplankton species by the taxonomers (Straile et al. 2013). If we are to integrate information of long-term datasets from various sources to study global scale phenomena, the challenge becomes even greater, as it is a labor intensive and costly project to achieve usable and reliable databases (Soranno et al. 2015). Given the fast technological development and the availability of numerous highly qualified people, it will not be long before we can efficiently harness the power of big data. But only nations with enough funding and personnel will have access to trustworthy big data. Research, not being the priority or even possibility of the majority of countries worldwide, is geographically biased. Take as an example how many scientific papers are available on the American vs. the African Great lakes. Even though the latter are equally significant for ecosystem services and biodiversity as the former, they are only studied sporadically by international projects (Bootsma 2018). Scientific advances that are heavily dependent on cutting-edge internet technology will probably perpetuate bias in our knowledge. In Chapter 6, we provide an opinion paper where we present several existing strategies for lake monitoring. We argue that it is hard to identify a monitoring strategy that works best for all purposes (scientific, policy making etc.) given the available funding, equipment, personnel and of course time and lake-accessibility. We present the strengths and drawbacks of the existing monitoring strategies and we provide an alternative method that could potentially increase research output at a wider geographic scale. Scientific Responsibility How can we move forward, without waiting for political activities to solve the world’s unbalanced distribution of wealth? By reinforcing grassroots initiatives where the power of the scientific outcome lies in the number of participants, regardless of their geographic origin. The European Multi Lake Survey (EMLS) was such an initiative. The EMLS took place in summer 2015 and consisted of 27 European countries, each with their own legislation and culture around the management of national water resources. The EMLS was inspired by the National Lakes Assessment of the Environmental Protection Agency in the USA (Pollard et al. 2017), that showed how to adequately sample numerous lakes over a short period. In Europe, although the Water Framework Directive stipulates guidelines to achieve good ecological condition, it neither covers costs for transnational surveys nor ensures the
11 participation of both EU and non-EU member states. For this reason, the organization of EMLS as a self-funded initiative was important. With this action, we showed that high-standard collaborative research could be carried through following the motto of “If you want to go fast, go alone; if you want to go far, go together”. Understanding the impact of global scale phenomena, requires information from many lakes of similar - and different - characteristics (e.g. morphometry, trophic status) across a large geographic scale, to demonstrate if they respond in a consistent manner to similar - and different - environmental forcing. In Chapter 6, we provide examples on how a space-for-time substitution approach can be used to study current spatial phenomena instead of long-term biotic records, and achieve such lake comparisons (Pickett 1989). Snapshot sampling, where lakes are sampled only once, is commonly used in multi-lake comparisons, such as we did in the European Multi Lake Survey. A predefined time-period is chosen, based on when the studied phenomenon is expected to happen, to avoid seasonality. Such sampling efforts produce comparable datasets, with uniform, synchronic data. Thus, we can obtain a valuable synoptic picture of the relationship between environmental predictors and biological responses that drive the structure of the phytoplankton communities at spatial scale. Space-for-time substitution can explain temporal patterns (Blois et al. 2013), but supporting studies are needed to fully understand the complex cause-effect relationships that aids cyanobacterial bloom occurrence and to predict future lake statuses. To turn the EMLS into a truly robust survey, we brought together lake experts, of different related disciplines, from two European COST Actions, CyanoCOST and NETLAKE). Collaboration and assistance was also achieved at a global level through participation in Global Lake Ecological Observatory Network. In Chapter 3, we describe the procedures we followed in order to ensure comparable datasets among data collectors. In brief, we established standardized sampling, sample processing and analyses protocols among lakes, during a three-day training school in Evian-Les- Bains, France (May 2015). We designed straightforward sampling protocols to accommodate the capacity in funding, available time, personnel and equipment of all participants, without compromising quality. Finally, we agreed on complete inclusiveness of all participants in three initial peer-reviewed publications. To achieve a large number of lakes for adequate spatial coverage, the representatives of each country reached out for more collaborations in their respective countries. As a result, 369 lakes spread across the continent (Figure 1) were sampled for chemical, physical and biological parameters. In this way, environmental gradients across wide geographic scales were covered with relatively little effort and with high cost efficiency. We sampled in summer as cyanobacterial blooms are a distinct feature of summer phytoplankton, during the locally warmest period, in order to test for temperature effects on cyanobacteria.
12 The shipping and storage of samples was centralized while lake samples for nutrients, algal pigments and cyanobacterial toxins were analyzed, in dedicated laboratories, by one person on one machine, to minimize variation in analytical errors. Quality control and integration of the different datasets did not require more than a month. The EMLS dataset has already benefitted several countries individually (Poland, Spain, Turkey), through national reports on cyanobacterial toxin profiles, national assemblies about water quality, and applications for national research funding, for example. The complete EMLS dataset and methodology have been accepted in Nature Scientific Data. The rest of the expected data (e.g. DNA analysis, flowcytometry) can be incorporated with the EMLS dataset in the online database of the Environmental Data Initiative (https://portal.edirepository.org/nis/mapbrowse?scope=edi&identifier=176&revision =4).
Figure 1 The 369 lakes of the European Multi Lake Survey that were sampled in summer 2015.
Research Questions The EMLS focuses on the importance of direct and indirect effects of global warming and eutrophication and the mode of interaction between them (Chapter 4). Although experiments and modelling studies clearly hint at a possible synergistic interaction between increased nutrients and temperature in promoting cyanobacterial blooms (Rigosi et al. 2014) and their toxin production (Lürling et al. 2017), convincing evidence from field observations is barely present. Water column stability (stratification) is also expected to increase as an indirect effect of increased temperature (Gerten and Adrian 2002). Enhanced stratification, may establish a
13 shallow near surface mixed layer restricting light availability within this narrow zone. The light availability will only favour phytoplankton that can maintain their position, such as buoyant cyanobacterial species (Reynolds, 2006), within this narrow zone. Thus, modified light climate (indirect effect of high nutrients) might also interact synergistically with enhanced stratification, accentuating cyanobacterial blooms at the continental scale. The hypotheses that we address in Chapter 4 are: 1. The direct effects of increased nutrients and temperature and their indirect effects – be it light climate and stratification strength, respectively – determine cyanobacterial and algal biomass and thus, phytoplankton community structure at the continental scale. 2. Synergistic interactions between the direct and indirect effects of nutrients and temperature contribute significantly to driving cyanobacterial and algal biomass. 3. Drivers and/or their interactions differ between climatic zones and lake depth types. The EMLS sampled mostly (hyper-) eutrophic lakes. Data analysis using interactive terms of nutrients and temperature can be helpful to determine if and how (synergistically, additively or antagonistically) the interaction of environmental stressors will determine total algal, cyanobacterial and green algal biomass across the continent. We test the effect of the aforementioned environmental variables on algal or cyanobacterial biomass between lakes of different depth type (shallow versus deep) or geographic location based on different climatic zones (Mediterranean versus Continental versus Boreal). Insight into how environmental change determines the production and distribution of cyanobacterial toxins (hepatotoxins, neurotoxins, cytotoxins) is necessary for risk assessment and management (Ibelings et al. 2014). Nevertheless, experimental studies are not yet conclusive in attributing toxin production to specific cyanobacterial species under specific environmental conditions (Neilan et al. 2013). Hence, the lack of consistency in experimental findings and in standardization of monitoring and sampling campaigns impedes understanding of how environmental stressors are linked to cyanobacterial toxin production and toxin quota (toxin concentration per unit algal biomass). In Chapter 5, we hypothesize that: 1. During a heatwave like in summer 2015, temperature, either through direct (surface temperature, epilimnetic temperature) or indirect (water stability expressed as maximum buoyancy frequency) effects, strongly influences the distribution of toxin concentrations and toxin quota. 2. Under high temperature stress, the stringent selection of specific well-adapted strains of cyanobacteria reduces toxin diversity, potentially promoting dominance by a few highly toxic variants. We use community ecology approaches to examine toxins as a “community” of potentially coexisting toxin types, instead of focusing on each toxin separately. Direct
14 and indirect effects of temperature and nutrients are tested to determine their significance in explaining the distribution of the different toxins at the continental scale. A toxin diversity index (TDI) is calculated, using the Shannon equation, and is tested against all sampled environmental variables to investigate how toxin mixtures will be shaped along the continent, and if environmental perturbation will potentially favour a few highly toxic strains. Higher toxin diversity could perhaps lead to higher stability in overall toxicity within a bloom, since if one toxin declines, another may increase, leading to persistence in overall toxicity. The overall conclusions are summarized and discussed in Chapter 7. While global warming continues and eutrophication persists, the direct and indirect effects of increased lake temperatures and nutrients will drive changes in cyanobacterial biomass and the distribution of cyanotoxins in Europe.
15 Acknowledgements The authors acknowledge COST Action ES 1105 “CYANOCOST – Cyanobacterial blooms and toxins in water resources: Occurrence impacts and management” and COST Action ES 1201 “NETLAKE – Networking Lake Observatories in Europe” for contributing to the realization of the European Multi lake Survey. Evanthia Mantzouki was supported by a grant from the Swiss State Secretariat for Education, Research and Innovation (SERI) to Bas Ibelings and by supplementary funding from University of Geneva. We would like to thank University of Geneva, University of Amsterdam, University of Wageningen and the German Environmental Protection Agency for providing financial and technical support for the analysis of nutrients, pigments and toxins.
16 CHAPTER 2 Understanding the key ecological traits of cyanobacteria as a basis for their management and control in changing lakes
E. Mantzouki, P. M. Visser, M. Bormans & B. W. Ibelings 2016, Aquatic Ecology
17 Abstract Anticipated climatic changes combined with eutrophication are predicted to enhance the dominance of several notorious cyanobacterial taxa. Cyanobacteria have many key ecological traits that may allow them to thrive under foreseen scenarios of environmental change. Understanding the eco-physiological traits of harmful species has proven important for their successful control and management. Indeed, if the links between key cyanobacterial traits and the specific environmental conditions that allow expression of these traits can be disrupted, we could identify (novel) means for operational control and mitigate or prevent water quality problems. A good example is artificial mixing of a lake that breaks down the water column stability on which fast floating, buoyant cyanobacteria depend. Based upon Reynolds’ functional phytoplankton classification, we focused on five groups of cyanobacteria that from a management point of view can be seen as homogeneous and have comparable environmental sensitivities. For each group, we present (i) its key traits, (ii) how these characteristics will maintain their function under future environmental change, (iii) explain how understanding the function of these traits can reveal the “Achilles heel” of the particular functional group, and (iv) which (combination of) control measures is most likely to be successful. Despite looking for specific environmental sensitivities of individual groups, we maintain that controlling nutrients remains the basis for managing blooms, no matter which functional type dominates. Providing further ecological knowledge to lake management could be the key to effective bloom control and healthier, sustainable freshwater ecosystems even in a warmer future. Keywords: climate change, cyanobacterial blooms, eutrophication, functional traits, lake management, microcystins
18 Introduction This publication is the first paper in a special issue of Aquatic Ecology on “Cyanobacterial blooms: Ecology, prevention, mitigation and control”. In this special issue, the various contributions discuss a range of control measures available to lake management to prevent and control nuisance blooms of cyanobacteria. In this contribution, we review what is known about those key traits of cyanobacteria, like buoyancy or nitrogen fixation that define their ecological niche and are key to their successful management. Our title, however, also speaks of “changing lakes”, and we cannot properly address cyanobacterial problems without discussing the – changing - environment they live in. So, this publication also outlines the main consequences of eutrophication and climate change, as two major drivers of bloom formation. Phosphorus and nitrogen pollution is the biggest threat to water quality worldwide (Schindler & Vallentyne 2008). Anthropogenic eutrophication from sewage, fertilizers and detergents is responsible for massive algal blooms that capitalize on the excess of nutrients. Cyanobacterial blooms are the most conspicuous sign of eutrophication, mostly resulting in a strong smell, discoloration of the water and accumulation of floating cyanobacteria at the lake surface (Ibelings and Chorus 2007). Highly eutrophic waters can support excessive cyanobacterial biomass that increases turbidity, limits the penetration of light to deeper layers and increases anoxia in the hypolimnion (Schindler et al. 2008). Managers have worked hard to control eutrophication. The International Lake Environment Committee`s (ILEC) review on 217 lakes showed that water management in industrialized areas succeeded in reducing nutrient input in many lakes (e.g. Lake Biwa – Japan, Lake Constance – Germany-Switzerland , Lake Balaton - Hungary, the Great Lakes in North America) but the nutrient concentrations did not yet return to the levels of the 1950s, before the peak in the eutrophication and in many lakes with high historical loads, internal loading will continue to fuel cyanobacterial blooms for many years after external nutrient inputs have been reduced. In many cases, like for instance Lake Veluwe - The Netherlands - lakes showed severe hysteresis to restoration measures and remained in a turbid state for decades, despite tackling external nutrient loading (Ibelings et al. 2007). An important question is which nutrient, phosphorus or nitrogen, is chiefly responsible to promote cyanobacterial biomass in an ecosystem. Schindler et al. (2008) proved experimentally that algal blooms in the eutrophic lakes they studied could not be controlled by reducing nitrogen input, but only by managing phosphorus. According to their whole-ecosystem experiment, reducing nitrogen inputs increasingly favored nitrogen-fixing cyanobacteria that outcompeted the other phytoplankton. Nitrogen fixation could sustain cyanobacterial biomass and the lake remained highly eutrophic, despite showing indications of extreme nitrogen limitation seasonally (Schindler et al. 2008). But this shift to nitrogen fixers does not occur in most N-limited lakes of low to moderate fertility and where it does, lakes typically remain N-limited despite the presence of nitrogen fixers (Paerl et al. 2011). N2 fixation can only sustain less than 50% of primary production, even when P
19 supplies are sufficient (Scott and McCarthy 2010). Furthermore, there are records of eutrophic systems dominated by non-N2 fixers that are N and P co-limited or even N limited (Paerl et al. 2011). Rigosi et al. (2014) showed that both nutrients are dominant factors promoting blooms. Thus, the feasibility of focusing only on P deficiency must be evaluated on a case-by-case basis (Lewis and Wurtsbaugh 2008) since it is likely that both N and P input controls are needed for long-term cyanobacterial bloom management. Indeed, we emphasize that nutrient control always needs to be the basis of lake restoration measures, aimed to reduce or prevent problems caused by algal blooms. There is mounting evidence that the current worldwide increase in cyanobacterial blooms is not just the result of anthropogenic eutrophication. Climate change, through direct and indirect effects of warming, is seen to play an increasing role in changing the properties of the aquatic ecosystems (Elliott 2012; Kosten et al. 2012; Paerl and Huisman 2008). Global warming has the potential to directly affect phytoplankton community composition as the growth rates of different phytoplankton taxa exhibit different optimal ranges (Butterwick et al. 2005). Apart from the direct consequences of warming, the increase in surface water temperature in perialpine lakes for instance has resulted in up to 20 % higher water column stability (Livingstone 2003). As a consequence of global warming, the thermocline of deep stratifying lakes remains stable for longer periods (earlier onset in spring, later and incomplete overturn in autumn) and becomes positioned at shallower depths, potentially enhancing the growth of certain cyanobacterial taxa (Paerl and Huisman 2008). Intensified vertical density stratification reduces vertical mixing resulting in enhanced scope for buoyancy controlled vertical positioning of cyanobacteria (Paerl et al. 2011). Under stable water column conditions, positively buoyant cyanobacteria will be able to regulate their position in the water column and remain in the euphotic zone while non-motile, negatively buoyant algae that can only remain suspended in the water column under mixing conditions will tend to sink (Reynolds 1984; Bormans and Condie 1998). Moreover, being closer to the lake surface buoyant cyanobacteria are closer to the source of CO2, diffusing in from the atmosphere (Paerl et al. 2011). Higher atmospheric carbon dioxide concentrations will result in higher dissolved carbon dioxide concentrations that may (partly) relieve the potential C-limitation of the cyanobacteria with higher cyanobacterial biomass as a consequence (Verspagen et al. 2014). Microelectrode studies, however, demonstrated that the local demand for carbon in cyanobacterial surface waterblooms far outstrips the supply, resulting in severe C-limitation, despite being at the surface (Ibelings and Maberly 1998). Furthermore, climate change may affect the occurrence of cyanobacterial blooms in other, perhaps more subtle ways. We give a few examples. Changes in precipitation are expected due to global warming (IPCC 2007). Intensified precipitation will result in higher terrestrial runoff, increasing nutrient input and stimulating blooms. Increased flushing rates on the other hand, may prevent blooms when lake residence times are reduced to very low values (Verspagen et al. 2006), but ultimately enhanced
20 nutrient loads always carry a risk of promoting blooms (Paerl and Huisman 2008). Droughts may reduce the flow rate of major rivers to values low enough to support cyanobacteria as found in many Australian rivers (Bormans et al. 2005, Viney et al. 2007), and this may increase their presence in the potamoplankton, even in temperate regions, as demonstrated for the rivers Rhine and Meuse in extremely warm and dry summers (Ibelings et al. 1998). According to the World Meteorological Organization’s 2013 report, the occurrence of some extreme events (e.g. hurricane Katrina in 2005, or floods in Pakistan in 2010) can be attributed to climate change (www.cgd.ucar.edu/cas/ace/). The impact of extreme events, like summer heat waves as occurred in Europe in 2003 on lake ecosystems is likely to be different from the effects of long-term gradual warming (Anneville et al. 2010; Gallina et al. 2011; Straile et al. 2010), not all summer heat waves for instance were shown to promote blooms of cyanobacteria (cf. Jöhnk et al. 2008 vs. Anneville et al. 2015). We have thus identified eutrophication and climate change as two major anthropogenic stressors that may favor cyanobacterial blooms. These two factors have a clear impact on algal blooms by themselves but they don’t necessarily work only independently. A high algal biomass in eutrophic systems, for instance, can increase the water temperature at a local scale due to strong light absorption (Ibelings et al. 2003). Consequently, eutrophication may promote warmer water and provide additional competitive advantage of buoyant over non-buoyant phytoplankton (Paerl and Huisman 2008; Figure 1). The underwater light climate as experienced by the phytoplankton is determined by two factors: light attenuation and mixing depth. Steep light attenuation in eutrophic lakes limits the availability of light to a shallow euphotic zone. Climate warming decreases the mixing depth shifting the advantage to buoyant cells that can float up into the illuminated near surface mixed layer (Ibelings et al. 1991; Figure 1). These surface blooms will further increase the underwater light extinction coefficient (Figure 1). These examples demonstrate the potential for eutrophication to ally with climate warming in promoting the conditions for cyanobacterial growth and dominance (Moss et al. 2011).
21
Figure 1 Conceptual figure, illustrating the environmental controls of cyanobacterial bloom dynamics, and the direct and indirect effects of climate change on these dynamics in combination with external nutrient loading. Figure taken from Paerl & Huisman 2009 with permission by Wiley Journals.
The general expectation is that on-going climate change will amplify eutrophication problems, but much uncertainty remains about the extent to which both stressors will interact. These two stressors may interact in different modes, additive, antagonistic or synergistic. Specifically, it remains an open question whether nutrient and temperature will interact synergistically, so that the combined effects of eutrophication and global warming will be worse than the effect of each single factor alone. Several authors, either through the use of long-term monitoring data or in modeling simulations, have studied the interaction between eutrophication and climate warming as important drivers of blooms (Elliott et al. 2005; Wagner and Adrian 2009). Studies by Kosten et al. (2012), Posch et al. (2012) and Paerl and Huisman (2008) point to higher temperatures combined with nutrients inputs as being important for cyanobacterial bloom formation. However, Jeppensen et al. (2005), using long-term data from 35 lakes located from the sub-tropics to the temperate zone in North America and Europe, argued that phytoplankton composition is primarily driven by nutrient loading, with global warming effects less detectable. Recent work by Rigosi et al. (2014) on ca. 1000 American lakes showed that both nutrients and temperature are main determinants of variation in cyanobacterial occurrence between
22 lakes, but they found no overall support for a significant synergistic interaction between nutrients and temperature on cyanobacterial biomass. Moreover, the relative importance of these two factors and their interaction was clearly dependent on lake trophic state and dominant cyanobacterial taxa. Apparently, there is no scientific consensus on the importance and mode of interaction between nutrients and temperature and further investigations are needed. Cyanobacterial blooms are very persistent and we need to work on better solutions to limit their occurrence and to control their impacts on lake ecosystems and the crucial ecosystem services they provide (Cardinale et al. 2011). Cyanobacteria and their key traits Ecology, as a predominantly biological discipline is concerned with the distributions of organisms and their interrelationships as well as the interactions with their environment (Reynolds et al. 2002). Phytoplankton possesses a high diversity of well- defined ecological traits that can vary predictably over different environmental gradients. Trait-based approaches in ecology demonstrate the importance of functional diversity in determining key properties of ecosystems and their functioning (Litchman and Klausmeier 2008). Typically, planktonic food webs are characterized by a high degree of redundancy indicating a strong functional similarity among species, with a small proportion of species exerting disproportionally strong effects on lake ecosystem functioning (Woodward 2009). Therefore, an important but not yet fully resolved question in the study of biodiversity and functioning of freshwater ecosystems is not what species are (species identity), but what species do (functional role) (Hillebrand and Matthiessen 2009; Suding et al. 2008). This implies that assemblages of organisms should be characterized based on phenotypic functional traits that respond directly to environmental changes (response-traits) and determine effects on aggregated processes at the community and ecosystem level (effect traits). Several studies have shown how key traits can favor the occurrence of trait-holders under defined sets of environmental conditions. This includes blooms of nuisance cyanobacteria, where decades of study have resulted in a fair understanding of how key traits determine the occurrence of the main bloom forming taxa (Dokulil & Teubner 2000). Therefore, lake management needs to explore ways to disrupt the conditions where cyanobacteria benefit from these traits, in order to better control bloom dynamics. Understanding the ecology of harmful species and their key traits proved already successful for their control and management in cases like the hypertrophic Lake Nieuwe Meer in the Netherlands (Visser et al. 1996), where artificial mixing removed the water column stability on which buoyant Microcystis depended. Ecology can provide the right guidelines for controlling nuisance harmful blooms and inform new technicological measures. Cyanobacteria, formerly known as blue-green algae, are the Earth’s oldest known oxygenic photoautotrophs and they have a global distribution (Cheung et al. 2013). Cyanobacteria are prokaryotic organisms that may occur as large colonies or filaments in lakes giving them a greater resistance to grazing (Reynolds 2006). They contain
23 chlorophyll a like all other phytoplankton taxa and specific accessory pigments, such as phycocyanin or phycoerythrin (Glazer 1984) that can help them to adapt to the available light conditions (de Marsac 1991). Cyanobacteria have several more eco- physiological adaptations that allow them to dominate aquatic systems under changing environmental conditions (Carey et al. 2012). For example, some cyanobacteria produce gas vesicles (Walsby 1994) that allow them to regulate their buoyancy (Ganf and Oliver 1982; Hudnell and Dortch 2008; Huisman et al. 2005). Under the increased thermally-stratified conditions, which are anticipated with global warming, these cyanobacterial taxa might be able to migrate between well- illuminated surface layers and nutrient-rich hypolimnetic waters (Ganf and Oliver 1982), escaping the increasingly nutrient-depleted epilimnia of lakes during extended stratification periods (Livingstone 2003). See, however, Bormans et al. (1999) reporting that no studies provide support for migration to sufficient depths to access nutrients in stratified systems. Cyanobacteria may also take direct advantage of warming because their growth rate will continue to increase with temperature; while the growth rates of many other phytoplankton taxa decline over 20°C (Litchman et al. 2010; Reynolds 2006). See Lürling et al. (2013), however, who found no significant differences in optimum growth temperatures between cyanobacteria and green algae. Other key traits such as the ability to fix nitrogen (Oliver and Ganf 2000; Reynolds 2006) and to produce dormant cells as a means to survive unfavorable conditions (Bartram and Chorus 1999) can provide cyanobacteria with a competitive advantage over other phytoplankton (Carey et al. 2012; Litchman et al. 2010), especially under increasingly unpredictable future climate conditions (IPCC 2007). See further discussion of these traits when we list the various cyanobacterial functional groups. Cyanobacteria are responsible for numerous problems with high environmental and socio-economical impacts. Cyanobacterial blooms, like all high algal crops, have considerable negative effects on aquatic food webs and ecosystem functioning by turning lakes anoxic, inducing fish mortality, reducing diversity and increasing habitat loss (Chorus and Bartram 1999; Havens 2008; Paerl et al. 2011). There is a growing concern that the frequency and magnitude of cyanobacterial blooms will be enhanced globally (Huber et al. 2012; Jöhnk et al. 2008), and that the geographic range of some cyanobacterial species will expand (Briand et al. 2004; Ryan et al. 2004; Sinha et al. 2012). Many cyanobacterial taxa produce various types of toxins, such as hepatotoxins, neurotoxins and dermatotoxins (Carmichael 1997). These toxins present a considerable risk to drinking water (Codd et al. 2005) and pose a substantial economic cost (Dodds et al. 2009; Steffensen 2008). The World Health Organization (WHO 1996) has been gathering information on toxic cyanobacteria including issues of human health, safe water practices, management, prevention and remediation. Toxic water blooms in many eutrophic to hyper-eutrophic freshwater ecosystems (sometimes also in low nutrient levels) worldwide are responsible for sporadic but recurrent episodes of wild and domestic animal illness and death as well as for occasional human poisonings from municipal and recreational water supplies
24 (Carmichael 2001). Toxic and non-toxic strains can coexist within a population (Kurmayer et al. 2002; Kardinaal et al. 2007; Briand et al. 2008). Experiments with Microcystis showed that increased temperatures independently and in interaction with elevated phosphorus concentrations can yield more toxic strains of Microcystis than non-toxic ones suggesting that nutrient loading and global warming may additively promote blooms with higher microcystin content (Davis et al. 2009). However, previous experiments with Planktothrix agardhii have shown that increased temperatures do not result in an increase of toxin production, while high nitrogen concentrations affected the increase of both toxic and non-toxic strains (Sivonen 1990). In these experiments though, the interaction between temperature and nutrients was not tested. Although cyanobacteria share certain key traits, they are a heterogeneous group, in which each taxon possesses different eco-physiological adaptations. For example, the size range of a cyanobacterial unit spans nearly five orders of magnitude (from the smallest single-celled cyanobacterial picoplankton to the multi-cellular filaments and colonies); the photosynthetic rates and growth rates of different taxa vary widely (Reynolds et al. 2002). Different combinations of these morphological and physiological traits might result in varied responses to changes in temperature and nutrients. Plasticity of these traits arises from differential gene expression within species and results in higher diversity of trait values (Litchman et al. 2010). Plastic responses reflected in morphological traits occur on a range of time-scales from sub- second (e.g. re-arrangements in the light harvesting antenna) to days (e.g. augmentation of the gas vesicle volume) in response to the prevailing environmental situation (Ibelings et al. 1994; de Tezanos Pinto and Litchman 2010a). While phenotypic plasticity can be advantageous under changing environmental conditions, it is dependent on a match between timescales of environmental fluctuations and plastic responses (Stomp et al. 2008). Thus, if an environmental disturbance occurs too fast, the cyanobacterial species present at the moment might not be as quick as necessary to adapt to the new conditions. This functional plasticity and its variation between cyanobacterial taxa have to be taken into account if we want to properly understand their behavior and achieve an effective control of toxic cyanobacterial blooms. Functional classification of cyanobacteria Trait-based approaches are gaining in popularity, given their opportunity to explain and predict group distributions along environmental gradients. It has been argued that patterns of functional diversity may afford better insights into processes of ecosystem change and responses (Reiss et al. 2009). Recognizing how key traits can enable the proliferation of certain species or functional groups can help to predict which cyanobacterial taxa are most probably going to increase disproportionally under eutrophication and global warming (Litchman et al. 2010). Trait-based phytoplankton ecology arguably started with Margalef (1987) (however also read Lund, 1964) who predicted the occurrence of different taxonomic/functional groups
25 of phytoplankton along gradients of nutrients and turbulence. A more detailed trait- based approach was developed by Reynolds (2002, 2006) who catalogued the natural assemblages of phytoplankton species, of contrasted water bodies with specific trophic status, according to their morphological and physiological key traits. Litchman and Klausmeier (2008) proposed the use of ecologically relevant traits of phytoplankton (morphology/cell size, temperature and reproductive related traits) as fundamental data to be integrated in mathematical models for predicting community structure under changing environmental conditions. Kruk et al. (2011) tested a morphology-based functional groups classification against environmental conditions, for a large dataset (211 lakes) covering a wide range of conditions while Fraisse et al. (2013) explained differences in phytoplankton community structure in rivers using a morphofunctional approach. Carey et al. (2012) gathered information about specific eco-physiological key cyanobacterial traits (ability to grow in higher temperatures; buoyancy; high affinity for and ability to store phosphorus; nitrogen-fixation; akinete production; efficient light harvesting) that are predicted to enhance cyanobacterial dominance under future climate scenarios. In this paper, we will use Reynolds (2006) functional classification, which is an update to the Reynolds et al. (2002) publication, specifically for the cyanobacterial taxa in there. We will focus on five selected functional groups that include taxa of concern such as Planktothrix agardhii, Cylindrospermopsis raciborskii, Anabaena flos-aquae (now Dolichospermum flos-aquae; Wacklin et al. 2009), Microcystis aeruginosa and Planktothrix rubescens, and show how trait-based cyanobacterial ecology can be used to our benefit for the management of these harmful blooms. Reynolds (2006) treated his phytoplankton species-counts as phytosociological relevés (small areas of uniform vegetation according to Tüxen et al. (1955) and diagnosed the species that were co-occurring frequently, rarely or not at all. The different clusters were categorized in an alphanumeric system that conveniently suits the patterns and periodic sequences of the natural assemblages in a given lake. Reynolds’ table (2006) (presented in a shortened and revised version in Table 1) enlisted the typical representatives of each particular phytoplankton association, which demonstrate similar morphologies and the habitat type where they are typically encountered. Additionally, the environmental tolerances and the related key functional traits of each group are enlisted providing an idea on what to expect in response to environmental change. For the purpose of this paper, the table identifies the environmental sensitivities of each group, which can in theory be exploited to compose dedicated management solutions (referred to by us as the “Achilles heel” of each group). In Table 1 we enlist only eight out of the 31 groups of Reynolds’ phytoplankton associations (those containing the major bloom forming taxa), from which several were combined to form three groups (S1/S2; H1/H2; LO/LM); additionally, we retain SN and R as separate groups. We combined the aforementioned associations into three groups, based upon the conviction that from a management/control point of view these can be treated as homogeneous groups, with comparable environmental
26 sensitivities. Further down we discuss the five groups in Table 1 by choosing one main representative per group, akin a holotype in taxonomy; that is of specific interest from an ecological and management point of view. For each group, we first discuss its key traits, then how the use of these key traits may be altered by environmental (climate) change, then if and how understanding the key traits opens up opportunities for the management and control of the functional group in question and finally which – combination – of control measures may actually be most likely to successful.
S1/S2: Turbid mixed layers – Planktothrix agardhii
Key traits. The representative species in S1/S2 (Table 1) are efficient light harvesters that have the ability to grow successfully under light limited conditions by adjusting their pigment composition effectively in response to the ambient irradiance (Fietz and Nicklisch 2002). In this paper, we focus on Planktothrix agardhii that dominates very turbid lakes. The populations of P. agardhii can be maintained throughout the year when the winter temperatures are mild (Sas and Ahlgren 1989). Filamentous cyanobacterial species such as P. agardhii are superior in light harvesting compared to larger colonial species, mostly due to morphological adaptations such as their slender, attenuated form (Reynolds 2006). P. agardhii requires little energy to maintain cell function and structure, thus it is capable of maintaining its biomass under light limiting conditions (Van Liere and Mur 1979). Planktothrix agardhii may produce microcystins (e.g. Tonk et al. 2005). Key traits under eutrophication and climate change. Climate warming can favor the dominance of P. agardhii since experiments have shown that its growth can increase steeply in a temperature range from 10 to 30 °C (Foy et al. 1976). Eutrophication on the other hand, stimulates dense blooms that reduce light availability in the water column and support P. agardhii populations (Carey et al. 2012). In highly enriched shallow waters, P. agardhii can outcompete other phytoplankton species – including other cyanobacteria - by diminishing the opportunities for light harvesting for those which are less well adapted to light deficiency (Reynolds 2006). Under nutrient rich conditions the excess of P. agardhii biomass promotes the conditions for its own growth by creating appropriate light limiting conditions (Reynolds 2006; Scheffer et al. 1997). Occurrence of P. agardhii blooms as an exponent of eutrophication in shallow lake ecosystems follows the concept of alternative stable states (Scheffer et al. 1997). At low P concentrations dominance of P. agardhii is absent, since there are insufficient nutrients to create the turbid conditions on which this cyanobacterium depends. With ongoing eutrophication, the conditions for P. agardhii gradually improve, until at a critical P-value the mixed phytoplankton community collapses and only P. agardhii remains. The water is now so turbid that only P. agardhii can maintain growth and sustain its population. Upon lake restoration (reduction in P-loading) initially nothing will happen, since also the turbid state is stable and maintains its own prolongation, helped for instance by high biomass of benthivorous fish species, like bream which stir up the sediment thereby increasing the turbidity too (Lammens et al. 2004; Roozen et al. 2003). A long period of hysteresis may follow (Ibelings et al. 2007). At some point
27 during the restoration process, however, the P concentration becomes too low to support a high P. agardhii biomass, in particular when external factors like cold winters break up the bloom (Berger 1975; Scheffer et al. 1997) or restoration measures like biomanipulation are carried out. At this point the underwater light climate has improved sufficiently to allow a return to a mixed phytoplankton community and P. agardhii will disappear. P. agardhii has a tendency to show biomodality in its distribution, either it is dominant or nearly absent, a hallmark of alternative stable states (Scheffer et al. 1997). Management actions. Controlling the nutrient input, thus improving the lake`s light environment by diminishing turbidity is the most fundamental solution to avoid the occurrence of filamentous cyanobacteria with efficient light harvesting abilities (Liere and Walsby 1982). However, several attempts to restore the original phytoplankton community in shallow lakes by reducing the nutrient loading were not always successful, in particular in the short term (Sas and Ahlgren 1989). Increased flushing rates could be effective for the management of P. agardhii. Observations from 27 case studies of lakes undergoing restoration (reduction of P loading) in Denmark demonstrated that in cases where the hydraulic retention time was long lakes recovered more slowly from cyanobacterial dominance than lakes with low retention times (<0.5 y), where the recovery was 10-300 times faster (Jeppesen et al. 1991). This can be explained by the fact that the maximum growth rate of the filamentous P. agardhii (as in many cases of several cyanobacterial taxa) is typically ca. only half of the maximum growth rate of green algae (r = 0.6 d-1 vs r = 1.2 d-1 respectively, calculated by Scheffer et al. 1997 based on Mur et al. 1977). Consequently, longer retention time favors P. agardhii while short retention time may remove it from the system, as P. agardhii no longer has the capacity to outcompete fast growing green algae. Scheffer’s et al. (1997) prediction model confirmed that P. agardhii would be absent independently of the nutrient level if the flushing rate was ≈18 % of the lake volume per day. Increased flushing rates along with increased water clarity can reduce hysteresis of lakes dominated by P. agardhii and lead to a faster return to a mixed phytoplankton community. Crucial for the long-term restoration of lakes formerly dominated by P. agardhii is the return of an abundant macrophyte community which competes for nutrients with the cyanobacterium, prevents sediment resuspension and provides shelter for large bodied zooplankton against fish predation (Gulati et al. 2008).
28 Table 1 The different cyanobacterial functional groups, with their preferable habitats, typical representatives, tolerances, key traits, sensitivities and the possible management actions are displayed.
In the S1/S2, H1/H2 and LO/LM combined groups; several habitat types and typical representatives are enlisted to facilitate the association of each group to certain key traits and possible management actions. The highlighted typical representative in each group is used further in the text to describe how certain key traits are affected by environmental changes and propose certain solutions for the management of their occurrence.
SN: Warm mixed layers- Cylindrospermopsis raciborskii
Key traits. According to Reynolds classification (Reynolds 2006), the main representatives of the SN group proliferate in warm mixed layers. In this paper, we will focus mainly on Cylindrospermopsis raciborskii whose occurrence is expanding. C. raciborskii blooms can negatively affect water quality since they can produce toxins such as the neurotoxin saxitoxin (Castro et al. 2004) and the alkaloid hepatotoxin, cylindrospermopsin (Li et al. 2001). The C. raciborskii distribution range used to be restricted to tropical and subtropical ecosystems, but since the 1930s it has progressively colonized more temperate environments, spreading from Greece and Hungary towards higher latitudes by the end of the 20th century (Padisák 1997). These expanding C. raciborskii strains have been reported as not toxic (Fastner et al. 2003; Yilmaz et al. 2008). However, a recent study by Gkelis and Zaoutsos (2014) revealed saxitoxin production in freshwater samples from Greece dominated by C. raciborskii. The expansion of C. raciborskii is supported by the existence of different physiological strains or ecotypes, which vary in their temperature tolerance with tropical strains to be tolerant in a temperature range of 15 to 35 °C (Chonudomkul et al. 2004).
29 C. raciborskii occurs mainly in shallow eutrophic waters with relatively high turbidity (Reynolds 2006). It holds common phenotypic traits with P. agardhii such as tolerance to continuous mixing and shaded water columns (Stüken et al. 2006). Another similarity is its high affinity for and ability to store phosphorous under limiting conditions (Istvánovics et al. 2000) that allows C. raciborskii to proliferate in warm surface mixed layers of stratified systems (Bormans et al. 2004). These shared key traits between P. agardhii and C. raciborskii, which are also depicted in their similar morphological characteristics, indicate functional equivalence and a probability of occupying comparable niches (Kruk et al. 2010). For example, in eutrophic lakes in Eastern Germany, cyanobacteria assemblages consist mainly of members of the S1 group (Reynolds 2006), but are often accompanied by C. raciborskii (Nixdorf et al. 2003). There are specific details however, which differentiate the niches in which each of these cyanobacterial species proliferate, for example C. raciborskii has higher light requirements for growth than P. agardhii (Briand et al. 2004). Additionally, C. raciborskii, as a Nostocales genus, has the capacity to fix atmospheric nitrogen in its heterocysts (Komárek and Mareš 2012).
Key traits under eutrophication and climate change. Experiments under different temperature regimes showed a positive net growth in a wide range of temperatures from 15 to 35 °C and light intensities ranging from 30 to 400 µmol photons m-2 s-1 of strains of both tropical and temperate origin (Briand et al. 2004). The colonization of C. raciborskii of mid-latitudinal environments hence may be a result of a high adaptability to a range of light and temperature gradients under perpetual environmental change induced by eutrophication and global warming (Briand et al. 2004). The characteristic of N2 fixation promotes its competitive strength under nitrogen-depleted environments, enabling it to effectively compete against other species like P. agardhii that is not able to fix nitrogen (Anagnostidis and Komárek 1988). Experiments, however, showed that C. raciborskii is capable to dominate over a mixed phytoplankton community of green algae and other cyanobacteria species regardless of N: P stoichiometry (Chislock et al. 2014); which reduces the capability to control its occurrence under eutrophic conditions in freshwater ecosystems. Moreover, C. raciborskii has the advantage of forming akinetes (Kaplan-Levy et al. 2010) as an effective means to survive unfavorable periods such as changes in light, nutrients, temperature or desiccation. These akinetes can also capitalize on the excess availability of P in the sediment upon germination (Istvanovics et al. 2000).
Management actions. Controlling the nutrient load of both N and P may be necessary to control the blooms of this harmful cyanobacterium (Chislock et al. 2014). Additionally, and similar to the previous group (S1/S2), Reynolds (2006) suggests that problems with C. raciborskii can be solved if the retention time in basins is shortened sufficiently under increased flushing rates. In a large impounded tropical river Bormans et al. (2005) showed that flushing was the only management option to terminate blooms of C. raciborskii.
30 H1/H2: Dinitrogen fixing nostocaleans – Dolichospermum flos- aquae (formerly Anabaena flos-aquae)
Key traits. In the H1/H2 group we have di-nitrogen fixing nostocaleans that occur mainly in small or large mesotrophic lakes (Table 1). Here, we will focus on Dolichospermum flos-aquae in order to show how the key traits that characterize the H1/H2 group support their tolerance to limiting nitrogen availability and how we can exploit their sensitivities to eliminate their presence in freshwater ecosystems. D. flos- aquae is a filamentous cyanobacterium that can be highly toxic when present in water resources (Agnihotri 2013) producing harmful neurotoxins such as anatoxin-a, anatoxin a(s) and homoanatoxin-a (Gallon et al. 1990). D. flos-aquae is a good competitor under low nitrogen supplies as it can fix nitrogen while it can escape poor light conditions by regulating its buoyancy (Kinsman et al. 1991; Reynolds 2006).
Nitrogen fixation is an energetically costly process that depends on light as an energy source. The rates of N2 fixation increase with increasing light levels (Agawin et al. 2007). Low N:P ratios stimulate heterocyst development (Agawin et al. 2007), resulting in an increase of the N-fixers’ biomass either in mesocosm experiments (Levine and Schindler 1999; Vrede et al. 2009) or in whole lake experiments (Schindler et al. 2008). Experimental studies also point out, that promotion of nitrogen-fixers only occurs when sufficient light is provided (de Tezanos Pinto and Litchman 2010a). Light regime has a primary significance for the success of D. flos-aquae as under low photon fluxes N:P ratios appear not to have a significant effect on its abundance (de Tezanos Pinto and Litchman 2010a). Under low light intensities, which stimulate gas vesicle production (Kinsman et al. 1991), D. flos-aquae may take advantage of its gas vesicles to regulate its buoyancy; float up and in this way may manage to enhance access to light and sustain its biomass (Carey et al. 2012).
Key traits under eutrophication and climate change. Anticipated changes in the environment can stimulate the occurrence of D. flos-aquae in different ways. Increased temperatures induce stronger stratification and shallower mixing depths, resulting in increased light availability for floating cyanobacteria like D. flos-aquae, which would benefit N2 fixation under low N availability. High temperatures also have a direct effect on optimizing N2-fixation by enhancing the rate of gas diffusion into the heterocyst (Bauersachs et al. 2014). Staal et al. (2003) found that Q10 (acceleration of a process over a 10 °C step, generally 10-20 °C) values for N2 fixation of heterocystous strains of Anabaena sp. were approximately equal to cyanobacterial growth rate responses to temperature (Q10≥1.8, Reynolds 2006).
Management actions. A potential remedy against blooms of N2 fixing cyanobacteria would be to reduce light availability. Deep mixing can have a negative effect on the growth of D. flos-aquae and its capacity to fix nitrogen, as it forces the population down to poorly illuminated waters with low energy supply. In general, N2 fixation is also restrained through the availability of phosphorus and micronutrients (molybdenum, iron) as the N2-fixing enzyme, nitrogenase, requires high levels of these nutrients to
31 operate (Vitousek et al. 2002). Hence, for several N2 fixers, Schindler (1977) puts forward that P is the ultimate limiting factor in freshwater ecosystems of moderate to high productivity, so that control of P could effectively control H1/H2 (see however Chislock et al. 2014 who showed that stoichiometric manipulation of N and P will not be enough as experiments on C. raciborskii showed its dominance under high or low N:P ratios). Reynolds (2006) in this case also points out that the H1/H2 group does not perform well under low phosphorus concentrations.
Lo /LM: Summer epilimnia - Microcystis aeruginosa
Key traits. Lo/LM are encountered when lake summer epilimnia are formed in mesotrophic and eutrophic lakes (Table 1). Our main concern here will be the genus Microcystis. Microcystis is a cosmopolitan genus which can generate dense cell accumulations (scums) on the surface of freshwater ecosystems, threatening human and ecosystem health (Visser et al. 2005). Several strains of Microcystis spp. are toxigenic producing a group of cyclic peptid hepatotoxins called microcystins and consist a risk for both public health and ecosystems (Carmichael 1997). Microcystis spp. forms large colonies that puts it at a disadvantage under nutrient limited conditions as its low surface area to volume ratio decelerates nutrient uptake rates (Reynolds 1997). Large colonies will move faster compared with smaller colonies or single filaments, following Stokes’s Law (Smayda 1970). Microcystis has been referred to as the champion of buoyancy regulation. Microcystis efficiently regulates its buoyancy by varying its dense cellular constituents (mainly sugars) on a diel time scale and buoyant colonies float up rapidly during periods of (partial) water column stability (Ibelings et al. 1991). Therefore, forming large colonies that can float fast is a key advantage for Microcystis (Ibelings et al. 1991). In experiments M. aeruginosa was able to migrate 12 m within 2h to gain access to both light and nutrients, defeating any density barriers (Ganf & Oliver 1982). Observations in the Dutch Lake Vinkeveen have demonstrated (Ibelings et al. 1991) that Microcystis spp. fast flotation velocity (Walsby 1991) allows it to efficiently track the near-surface mixed layer (Humphries and Lyne 1988) floating up until it reaches the well mixed layer near the lake surface where even large colonies will remain entrained in turbulent flow.
Key traits under eutrophication and climate change. Increased temperatures are positively related to M. aeruginosa’s growth rate as it reaches a maximum replication rate at 28 °C while its Q10 – here the increase in growth rate with a 10 o C increase in temperature –reputedly is close to 9.6, one of the highest value recorded in prokaryotic and eukaryotic phytoplankton species (Reynolds 2006, Carey et al. 2012). However, Lürling et al. (2013) demonstrated experimentally that the optimum growth temperature for several cyanobacterial and green algal taxa was not significantly different. Thus although, Lürling et al. (2013) suggest that M. aeruginosa might not benefit from higher water temperatures, Paerl and Huisman (2009) do support the notion that M. aeruginosa’s blooms will be promoted albeit through the indirect effects of climate warming. Increased temperatures lead to intensified density differences
32 enhancing lake stratification. When the stratification is well developed, then shallow wind induced mixing in the surface layer cannot be transferred to deeper layers, denying other phytoplankton the opportunity to become entrained in turbulent flow. Under stable stratification and reduced turbulence, M. aeruginosa colonies will be larger and their buoyancy will be further enhanced during vertical migration cycles (O’Brien et al. 2004). Consequently, M. aeruginosa’s ability to efficiently control its movement offers a clear advantage over non-migrating phytoplankton species under the expected prolonged and enhanced stratification (Carey et al. 2012, Paerl and Huisman 2009).
Management actions. Reynolds (2006) proposes a prolonged deep mixing regime in order to control the occurrence of M. aeruginosa (Table 1). Management of Microcystis in the Dutch Lake Nieuwe Meer showed that artificial mixing using bubble plumes can effectively prevent nuisance blooms and promote a mixed community of flagellates, green algae and diatoms (Visser et al. 1996). Non-buoyant algae were favored since their sedimentation rates were reduced due to the artificial mixing while the turbulent flow diminished the advantage of M. aeruginosa to float up into the illuminated near surface mixed layer (Visser et al. 1996). Under mixing, cells face rapid changes in light regime and competition leans towards those eukaryotic competitors that are better adapted to fluctuating irradiance levels compared to M. aeruginosa (Ibelings et al. 1994).
R: Metalimnia in stratified lakes - Planktothrix rubescens
Key traits. The representatives of group R (Table 1) are encountered in the metalimnion of mesotrophic, stratified lakes (Reynolds 2006). Planktothrix rubescens is a filamentous cyanobacterium that may produce the hepatotoxin microcystin (Anagnostidis and Komarek 1988). It contains gas vesicles that allow it to adjust its buoyancy and regulate its position in the water column. In contrast to Microcystis, however, it positions its populations away from the saturated irradiance levels at the lake surface (Jacquet et al. 2005). It belongs to the same genus as P. agardhii from the S1/S2 group, and both species are able to tolerate low light intensities as both share the same slender filamentous form and auxiliary pigmentation (Reynolds 1997). Nevertheless, the niches of the two congeners are clearly different. Whereas P. agardhii prefers a shady (turbid) environment, P. rubescens requires a relatively clear epilimnion. P. rubescens forms metalimnetic layers in deep, stratified meso-eutrophic lakes where it can completely monopolize resources and dominate the phytoplankton community (Anneville et al. 2004). At metalimnetic depths, the light quantity and light quality is different compared to the upper mixed layers and P. rubescens is well adapted to these light conditions due to the abundance of the red pigment phycoerythrin.
The morphology of gas vesicles has evolved to withstand the combined pressures in the lake environment (Walsby, 1994). Strains of P. rubescens have narrow gas vesicles (diameter = 50 nm) with critical collapse pressure (pc) ≈ 1.1 MPa compared to M. aeruginosa (d=65 nm, pc ≈ 0.85 MPa) as response to withstand increased hydrostatic
33 pressure that comes with increased depth (Visser et al. 2005). The P. rubescens gas vesicles will survive during deep winter mixing and cells maintain their buoyancy (Carey et al. 2012). Moreover, following Stoke’s law in non-turbulent conditions, filaments with a small diameter will have a low sinking or flotation velocity, compared to large colony forming taxa (Walsby 2005) and will find it easier to maintain position in the metalimnion. The intimate interactions between carbohydrate accumulation and gas vesicle-mediated buoyancy will support close to neutral buoyancy and will maintain filaments at a given depth (Walsby et al. 2004). Overall, P. rubescens is able to adjust the depth of its stratifying population in response to changes in the underwater light regime, and will take up different positions in the metalimnion if this is to its advantage (Jacquet et al. 2005).
Key traits under eutrophication and climate change. P. rubescens is present in high biomass during summer, when the phosphorus concentrations are very low in the epilimnion, but still remain available in deeper layers of lakes (Anneville et al. 2002). P. rubescens can even keep developing during autumn when there is lower irradiance due to increased mixing, forcing the thermocline to greater depth (Anneville et al. 2004). P. rubescens can be successful under different trophic states of the same lake; this has been observed in Lake Zurich (Switzerland) where P. rubescens has been found in phytoplankton assemblages during both the eutrophic and the meso-eutrophic stages of the lake (Anneville et al. 2004). In Lake Bourget (France), however, the long-term dynamics show a different pattern. P. rubescens was absent prior to eutrophication in the 1950s, while also the very nutrient rich conditions during the peak of the eutrophication prevented proliferation of this filamentous species. P. rubescens came to dominate Lake Bourget in the mid-1990s when lake restoration measures began to take hold. Lake restoration has been successful, even to the point that since 2009 P. rubescens once more is absent from the lake. Hence the relationship between trophic state and proliferation of P. rubescens follows a hump shaped pattern, where the peak of the hump is positioned at mesotrophic to meso-eutrophic conditions (Dokulil and Teubner 2012).
Recent studies show the importance of the seasonal effects of climate change on the deep, meso-eutrophic lakes in the peri-alpine region where P. rubescens typically dominates (Anneville et al. in press; Posch et al. 2012). Contrary to the common belief that blooms prefer hot summers (Jöhnk et al. 2008) extremely warm summers (like the one in 2003) do not promote blooms of P. rubescens, to the contrary they may reduce its population size. Apparently, P. rubescens does not benefit from extremely long and stable stratification during heatwave summers, despite the ongoing depletion of epilimnetic nutrient concentrations. Warm conditions in autumn and winter however, do promote P. rubescens in the following year, indicating the vital role of carrying on a large enough inoculum for growth through the winter to the next spring period.
Management actions. The return of P. rubescens in Lake Bourget in the mid 1990’s upon a return to mesotrophic conditions is attributed to increased transparency following
34 the restriction of other phytoplankton species under increasingly reduced phosphorus availability (Jacquet et al. 2005). Ernst et al. (2009) suggested that the effects of climate change strengthen the dominance of P. rubescens in peri-alpine lakes that undergo re- oligotrophication, because this cyanobacterium can take advantage of the earlier onset of stratification. Reduced mixing in warmer winters (incomplete overturn of the water column) will stimulate survival of intact gas-vesicles in P. rubescens (Walsby et al. 1998), aiding to achieve a large size of the inoculum. For these reasons, Reynolds suggested to alter the stability of the water column as an effective means to disrupt the development of P. rubescens (Reynold 2006).
Lake managers worldwide, for instance in the Netherlands, are aware of the increasing risks of toxic cyanobacteria during heatwaves, resulting in lakes being closed for recreation on an extensive scale (Ibelings et al. 2014). In contrast to the peri- alpine region however, lakes in the Netherlands tend to be dominated by cyanobacteria from the Lo/LM group. These “subtle” differences between effects of warming in summer vs. winter upon different functional groups - R vs Lo/LM - underlines once more the necessity to study and control cyanobacterial blooms upon the right – ecological – basis. Cyanobacteria with different key traits will respond differently to changes in the environment and require a different management approach.
Conclusions and research perspectives Cyanobacterial blooms are still a major problem in freshwater ecosystems in spite of major nutrient reductions over the last 40 years. Lake restoration has had mixed success in controlling their development. The anticipated changes of our climate have the capacity to worsen the problem, demanding appropriate and timely measures from lake managers. In this review, we demonstrate how management actions should be tailored depending on the dominant cyanobacterial functional group. Identifying the key traits of dominant cyanobacteria and the environmental conditions under which these traits succeed can reveal the Achilles heel of cyanobacterial functional groups, and provide lake managers with the most appropriate tools for their effective management and control. A comprehensive view of the many adaptive strategies of cyanobacteria to a wide range of environmental conditions is essential as controlling one species has also the potential to let another (maybe even more problematic) species proliferate. This knowledge on the cyanobacterial key traits should be more incorporated in various models developed and used to predict bloom occurrences. Yet, crucially all taxa ultimately can and should be controlled by managing the external and internal nutrient loading of lakes. This special issue in Aquatic Ecology presents the state of the art in cyanobacterial bloom management and control, continuously keeping the key ecophysiological traits of the culprits in mind.
35 Acknowledgments We would like to acknowledge two EU COST Actions; ES 1105 “CYANOCOST - Cyanobacterial blooms and toxins in water resources: Occurrence, impacts and management” and ES1201 “NETLAKE - Networking Lake Observatories in Europe” that offer us the possibility to develop the idea of this manuscript through numerous discussions with experts and researchers on cyanobacterial blooms. We would also like to thank all the anonymous reviewers for helping us improve our manuscript with all their constructive comments.
36 CHAPTER 3 A European Multi Lake Survey dataset of environmental variables, phytoplankton pigments and cyanotoxins
Mantzouki et al. 2018, Nature Sci Data
37 Abstract Under ongoing climate change and increasing anthropogenic activity, which continuously challenge ecosystem resilience, an in-depth understanding of ecological processes is urgently needed. Lakes, as providers of numerous ecosystem services, face multiple stressors that threaten their functioning. Harmful cyanobacterial blooms are a persistent problem resulting from nutrient pollution and climate-change induced stressors, like poor transparency, increased water temperature and enhanced stratification. Consistency in data collection and analysis methods is necessary to achieve fully comparable datasets and for statistical validity, avoiding issues linked to disparate data sources. The European Multi Lake Survey (EMLS) in summer 2015 was an initiative among scientists from 27 countries to collect and analyse lake physical, chemical and biological variables in a fully standardized manner. This database includes in-situ lake variables along with nutrient, pigment and cyanotoxin data of 369 lakes in Europe, which were centrally analyzed in dedicated laboratories. Publishing the EMLS methods and dataset might inspire similar initiatives to study across large geographic areas that will contribute to better understanding lake responses in a changing environment. Background and summary Eutrophication still is the primary process threatening lakes and reservoirs and the services they provide, like good quality drinking water, irrigation, fisheries and recreational opportunities. Anthropogenic eutrophication is responsible for massive algal blooms. Cyanobacteria have diverse functional traits that allow them to proliferate under various environmental conditions (Carey et al. 2012). The frequency and size of cyanobacterial blooms is increasing globally (Huber et al. 2012; Sinha et al. 2012). Excessive cyanobacterial biomass reduces light penetration and enhances anoxia in the hypolimnion, thus reducing species habitats and biodiversity (Schindler & Vallentyne, 2008). Moreover, the toxins produced by bloom-forming cyanobacteria present a considerable risk to drinking water (Codd et al. 2017) and pose a substantial economic cost (Dodds et al 2009; Steffensen 2008).
Climate change, through direct and indirect effects of warming, increasingly plays a role in changing physico-chemical and biological properties of aquatic ecosystems (Paerl & Huisman 2008; Elliot, 2012; Kosten et al. 2012; Schwefel et al. 2016), contributing to the global increase in cyanobacterial blooms. Although optimal growth temperatures vary widely between cyanobacterial strains and species, as well as for their eukaryotic competitors, their growth rate increases faster with temperature than for other phytoplankton groups (Reynolds 2006; Kosten et al. 2012). Longer ice- free seasons, reduced winter overturn and enhanced water column stability in summer may all indirectly favour cyanobacterial blooms. In deep peri-alpine lakes for instance, water column stability has increased by 20% in a response to warming of the atmosphere (Livingstone 2003).
38 Interactions between nutrients and temperature-related changes are expected (Moss, 2011; Rigosi et al. 2014; Lürling et al. 2017). However, it is still uncertain to what extent and following which mechanisms nutrients and temperature will interact to amplify blooms. Climate forcing of blooms will differ among regions (Salmaso et al 2013). For example, at high latitudes and equatorial areas, intense precipitation events are expected (IPCC7) to increase nutrient enrichment of water bodies from enhanced surface runoff and groundwater discharge (Schindler & Vallentyne, 2008), whereas at low and mid-latitude continental interiors, droughts (Dai et al. 2004) may reduce river flow rates, increase lake residence times and thereby promote cyanobacteria. In hyper- eutrophic systems, high algal biomass may even increase local temperature due to enhanced light absorption (Ibelings et al. 2003). Thus, high nutrient concentrations may promote warmer temperatures, giving a competitive advantage to buoyant cyanobacteria over non-buoyant algal species (Schindler & Vallentyne, 2008). All of these interactions between nutrients and temperature may vary with lake depth. Shallow lakes respond more directly to nutrients and temperature than deep lakes, which, in contrast, are more typically subjected to the indirect effects of the aforementioned drivers (Gerten & Adrian, 2002).
Hence, a complex interplay of regional (climate), local (nutrients, lake morphometry) and biological (species, functional groups) lake variables determines cyanobacterial bloom formation. Consequently, studies covering different regions and lake types are needed to disentangle the relative importance of the environmental predictors and their interactions. The European Multi Lake Survey (EMLS) obtained a deeper insight into cyanobacterial dynamics under different ecosystem variables across Europe. A space-for-time substitution, where contemporary spatial phenomena are studied in many lakes, instead of long-term temporal studies in a limited number of lakes (Pickett, 1989), was used. The survey took place in summer 2015, Europe’s third hottest summer on record, and comprised scientists from 27 countries that sampled 369 lakes only once. In this way, environmental gradients across wide geographic scales were covered with relatively little effort and with higher cost efficiency. We sampled in summer as cyanobacterial blooms are a distinct feature of summer phytoplankton (Sommer et al. 2012), during the locally warmest period, in order to test for temperature effects on cyanobacteria. In EMLS, standardized sampling procedures were strictly followed to ensure data homogeneity and eliminate site or operator related observation effects. Finally, lake samples for nutrients, algal pigments and toxins were analyzed in dedicated laboratories by one person on one machine, minimizing variation in analytical errors. Apart from providing a solid research dataset on which several analyses are being conducted, the EMLS helped to enhance the standards for limnological data collection and stimulate international collaboration. A subset of the EMLS toxin dataset has already been used in a recent publication to show how the distribution in cyanobacterial toxins and toxin quota was determined by both direct and indirect
39 effects of temperature (Mantzouki et al. 2018). Here we additionally present the data for all lakes, including those without toxins for the full set of data. The intention is that this publication of the EMLS dataset will further demonstrate the feasibility and value of snapshot surveys, and encourage similar programs in a continuously changing environment, i.e. at times when datasets covering large geographical gradients are in great demand.
Figure 1 Map showing the locations of the 369 lakes sampled during the European Multi Lake Survey in summer 2015.
Methods Organization To make EMLS a robust survey we bridged two European COST Actions (COST, 2017), CyanoCOST (Cyanobacterial blooms and toxins in water resources: occurrence impacts and management - CYANOCOST, 2014; Meriluoto et al. 2017) and NETLAKE (Networking Lake Observatories in Europe - NETLAKE, 2016), planting the idea and promoting the benefits of an extended collaboration amongst researchers from all over Europe. This research was expanded to many other scientists not directly involved in these COST Actions. The EMLS protocols required that each group of data providers collected and handled the samples following the same standards. Therefore, the steps outlined below had to be followed by all participant groups to ensure standardized sampling, sample processing and analyses, resulting in a homogenous dataset. Any deviations from
40 these protocols were recorded in the metadata spreadsheets, and these data were handled with care after contacting the data collectors. To ensure that protocols were fit-for-purpose and understandable to everyone, we invited representatives of each country involved in the CyanoCOST and NETLAKE actions to a three-day training workshop in Evian-Les-Bains, France. During the workshop, participants discussed all aspects of EMLS and considered limitations in the financial and logistic means for given countries, without compromising research quality. They finalized the protocols and obtained hands-on experience in using them. To increase the number of studied lakes and achieve adequate spatial variation, representatives of each country acted as EMLS-ambassadors, reaching-out to further collaborators within their own countries and disseminating the decisions and protocols of the EMLS. The EMLS was a collective effort, which means that each participating group used their own financial means to conduct their sampling as well as provided the personnel and facilities needed. Since the EMLS was a zero-funding effort, individual countries mainly contributed with samples from lakes that they routinely sample anyway, especially lakes with a history of eutrophication, given the implications for lake management. Although this results in a bias towards productive lakes, unfortunately it also reflects reality, with many lakes in Europe still suffering from eutrophication. A total of 369 lakes were sampled, spanning from Cyprus to Finland, and from the Asian part of Turkey to the Portuguese Azores islands (Figure 1). The workflow for organizing the EMLS consortium is presented as an infographic (Figure 2). It illustrates the logistics from organizing the local surveys (1) through obtaining the samples (2), to processing and shipping them to the analytical laboratories (3-5). The data obtained from the field as well as laboratory-analyses was quality controlled and integrated into a unique dataset before making available to the EMLS network and to the rest of the scientific community. The methods of data acquisition and laboratory analyses that are described below are expanded versions of descriptions in our related work (Mantzouki et al. 2018).
41
Figure 2 Schematic overview of the European Multi Lake Survey (EMLS). CyanoCOST and NETLAKE members performed the sampling routine by: 1. Preparing the sampling material to meet the standardized procedures 2. Accessing the lake site and acquiring integrated water samples from the bottom of the thermocline up to the water surface 3. Processing and preserving the water samples based on requirements for subsequent analysis 4. Shipping samples to a central receiving laboratory 5. Analysing nutrient, pigment and toxin concentrations in dedicated laboratories. After quality control checks and data integration the dataset returns to the European network and becomes publicly available for further research. Created by Sarah O’Leary and Eilish Beirne.
42 Data acquisition Date and location (in situ) The field methods for the EMLS were designed to be completed within one field day for each lake, but typically the sampling itself could be completed within two hours. Remote, poorly accessible lakes required more time to reach and field crews had to plan accordingly. To optimize time investment, lakes in close proximity were covered in one sampling trip. Each sampling group was responsible to organize and prepare the sampling material and equipment for their sampling campaign. In several cases, sampling groups of different areas or even countries collaborated and shared material such as instruments and boats. Each EMLS data collector team had to identify the right sampling period, defined as the warmest two-week period in summer, based on long-term air temperature data in each region, covering the last 10 or more years. This predefined time-period served to minimize confounding effects of seasonality. Each lake was sampled within this time window (Date; Table 1). The sampling location for each lake was defined as the central point of the lake. If a particular lake had been previously sampled at a specific location for long-term monitoring, the sample location from the long-term monitoring was used instead of the lake centre. For lakes with more than one relatively isolated basin, individual basins were sampled separately when possible, and indicated as such in the dataset (e.g. TR_BEY_I and TR_BEY_II). The latitude and longitude of the sampling location were recorded with a GPS device and provided in decimal degrees according to the WGS84 coordinate system. If cyanobacterial surface blooms (defined as the presence of a visible surface scum) were present close to where the team entered the lake or in close proximity to the sampling point, a second sampling location was considered. A scoop surface sample, using a small sealable container like a Falcon tube, of the cyanobacterial scum was acquired along with the water column sample. The location of the scum sample was noted. Data collectors also provided - when available – maximum depth, mean depth and altitude (Table 1) of the lake. Temperature and Secchi depth (in situ) Temperature profiles were measured with available probes, such as various CTDs or a Fluoroprobe (BBE Moldaenke). If no profiling instrument was available, water samples were taken at 0.5 m intervals from the lake surface to the bottom of the thermocline (top of the hypolimnion) and water temperature was measured using hand-held thermometers directly after sampling. Here, the thermocline range was defined as the depth interval at which the rate of temperature decreased at least 1 °C per metre. The bottom of the thermocline (where temperature no longer decreased by 1 °C per meter), defined the sampling depth in the case of stratified lakes. For lakes where a thermocline was not observed (mostly shallow lakes), the temperature profile
43 was measured until the lake bottom. In this case, the sampling depth was determined at 0.5 m above the lake bottom. From the temperature profiles, we obtained surface and epilimnetic temperature (Table 1) as the average temperature from surface until the bottom of the thermocline. The temperature profiles were also used during data analysis, to calculate the location where the thermocline lies even between two temperature measurement depths, which corresponds to where the thermocline is the most stable (point of maximum buoyancy frequency). To calculate this thermocline depth (ThermoclineDepth_m, Table 1) we used the command thermo.depth from the R package rLakeAnalyzer28. The Secchi depth (Table 1) was also recorded using a Secchi disk to the nearest 0.05 m. Integrated water sample (in situ) All data collectors constructed a simple device, known as the “Anaconda”, using a stoppered hose of the correct length in order to acquire the epilimnetic sample. The hosepipe was lowered with the bottom end open into the water column until the right depth (see above). When the hosepipe was vertical and the water level was visible at the surface layer of the hosepipe then the stopper was inserted to create hydrostatic pressure. The bottom end of the hosepipe was pulled-up with a rope to the surface to collect the sample in a bucket. The diameter of the hosepipe was appropriate to sample the required water volume (about 5-10 L for hypertrophic and eutrophic, 15-30 L for mesotrophic and oligotrophic lakes) for the analyses, in an acceptable number of runs. The first three sampling runs served the purpose of rinsing the hosepipe, the sampling bucket and the plastic rod. The subsequent runs were the water sample taken for analysis. The water sample in the bucket was mixed adequately before being divided into different bottles for further processing prior to analysis. All samples were shipped frozen using dry ice in Styrofoam boxes. Shipping and storage of the EMLS samples was centralized at the University of Wageningen (The Netherlands). There, samples were sorted and sent to the dedicated laboratories for further analysis. Each of the nutrients, pigments and toxins analyses were done in one dedicated laboratory, by one operator on one machine, to minimize analytical errors and maximize integration of the datasets. Specifically, the nutrients, microcystins and nodularin analyses were done at the University of Wageningen, the pigment analysis at the University of Amsterdam and the cylindrospermopsin and anatoxin analysis at the German Environment Agency. Total and Dissolved Nutrients (laboratory) For analyses of total phosphorus (TP_mgL, Table 1) and nitrogen (TN_mgL, Table 1), unfiltered water subsamples of 50 mL were stored in -20 °C until shipping. For dissolved nutrients: orthophosphate (PO4_ugL, Table 1), nitrite/nitrate (NO3NO2_mgL, Table 1) and ammonium (NH4_mgL, Table 1), a volume of 250 mL was filtered through 47mm Glass fibre filters (GF/C or GF/F or similar), the filtrate was sampled in a PE bottle and stored at -20 °C. Before the collection of the nutrient samples, all polyethylene collection bottles with their screw caps were acid washed
44 overnight in 1M HCl and rinsed with demineralized water and lake water before collection. Nutrients were measured according to Dutch NEN standards, using a Skalar SAN+ segmented flow analyser (Skalar Analytical BV, Breda, NL) with UV/persulfate digestion integrated in the system. The total phosphorus and orthophosphate were analyzed conforming NEN (1986), the ammonium and total nitrogen according to NEN (1990) and the nitrite/nitrate following NEN (1997). The limit of detection was 0.02 mg/L for total phosphorus and ammonium, 0.2 mg/L for total nitrogen, 0.01 mg/L for nitrite/nitrate and 0.004 mg/L for orthophosphate. Pigment analysis (laboratory) For pigment analysis, a volume of 50-250 mL for hypertrophic and eutrophic lakes and 500-1000 mL for mesotrophic to oligotrophic lakes was filtered through 47mm glass fibre filters (GF/C or GF/F or similar) using a filtration device. Filters were stored at -20 °C in the dark until shipping. The analysis of pigments was modified from the method described by Van der Staay et al. (1992). All filters were freeze dried for 6 hours. Filters were cut in half, placed in separate Eppendorf tubes, and kept on ice until the end of the extraction procedure. In each tube, 600 µl of 90 % acetone were added with a small amount of 0.5 mm glass beads. To extract the pigments from the phytoplankton cells, filters were placed on a bead-beater for one minute. To increase the extraction yields, samples were placed in an ultrasonic bath for ten minutes. This procedure was repeated twice to ensure a complete extraction of the total pigment content of the filters. To achieve binding of the pigments during the High- Performance Liquid Chromatography (HPLC) analysis, 300 µl of a Tributyl Ammonium Acetate (1.5 %) and Ammonium Acetate (7.7 %) mix were added to each tube. Lastly, samples were centrifuged at 15, 000 rpm at 4 °C for ten minutes. 35 µl of the supernatant from both Eppendorf tubes of a filter were transferred into an HPLC glass vials. Pigments were separated on a Thermo Scientific ODS Hypersil column (250 mm × 3 mm, particle size 5 μm) in a Shimadzu HPLC machine and using a KONTRON SPD-M2OA diode array detector. We identified 12 different pigments (chlorophyll-a, chlorophyll-b, zeaxanthin, diadinoxanthin, fucoxanthin, diatoxanthin, alloxanthin, peridinin, chlorophyll-c2, echinenone, lutein and violaxanthin, Table 1) based on their retention time and absorption spectrum and quantified by means of pigment standards. The limit of detection (LOD) and for a 250 mL sample was: 0.094 µg/L for chlorophyll-a, 0.061 µg/L for chlorophyll-b, 0.034 µg/L for zeaxanthin, 0.053 µg/L for diadinoxanthin, 0.067 µg/L for fucoxanthin, 0.029 µg/L for diatoxanthin, 0.023 µg/L for alloxanthin, 0.085 µg/L for peridinin, 0.028 µg/L for chlorophyll-c2, 0.031 µg/L for echinenone, 0.027 µg/L for lutein and 0.075 µg/L for violaxanthin. The limit of quantification (LOQ) and for a 250 mL sample was: 0.315 µg/L for chlorophyll-a, 0.202 µg/L for chlorophyll-b, 0.115 µg/L for zeaxanthin, 0.177 µg/L for diadinoxanthin, 0.224 µg/L for fucoxanthin, 0.098 µg/L for diatoxanthin, 0.077 µg/L for alloxanthin, 0.284 µg/L for peridinin, 0.093 µg/L for chlorophyll-c2, 0.104 µg/L for echinenone, 0.089 µg/L for lutein and 0.250 µg/L for violaxanthin. In the cases where no pigment signal was detected, the respective pigment was considered absent and noted as 0
45 μg/L in the dataset. If the calculated pigment concentration in the dataset is above the limit of detection (qualitatively detected signal) but below the quantification limit (too small to quantify), we suggest the assignment of a very small value of half the detection limit to enable the inclusion of these samples in statistical analyses (if applicable). Alternatively, other statistical approaches that account for data censoring can be followed based on the research question and the statistical analysis followed (for suggestions see33). Cyanotoxin analysis (laboratory) For toxin analyses, a volume of 50-250 mL for hypertrophic and eutrophic lakes and 500-1000 mL for mesotrophic to oligotrophic lakes, was filtered through 47 mm Glass fibre filters (GF/C or GF/F or similar) using a filtration device. Filters were stored at - 20 °C until shipping. In the laboratory, frozen filters were transferred to 8 mL glass tubes and freeze-dried (Alpha 1-2 LD, Martin Christ Gefriertrocknungsanlagen GmbH, Osterode am Harz, Germany). The freeze-dried filters were used for the Liquid Chromatography with tandem Mass Spectrometry detection (LC-MS/MS) analysis of microcystins, nodularin, cylindrospermopsin and anatoxin as described below. In the cases where no toxin signal was detected, the respective toxin was considered absent and noted as 0 μg/L in the dataset. A similar approach as in the section “pigment analysis” can be followed for toxin concentrations in the dataset that fall above the detection limit but below the quantification limit, as we did in (Mantzouki et al. 2018). Microcystins and nodularin analysis (laboratory) For the extraction of microcystins and nodularin, 2.5 mL of 75 % hot methanol – 25 % ultrapure water (v/v) was added to the freeze-dried filters, which were then sealed with a screw cap and placed for half an hour at 60 °C. Subsequently, the extract was transferred to a clean 8 mL glass tube. This extraction procedure was performed three times for each filter. The supernatants of the repeated extraction procedure were combined to a final volume of 7.5 mL and then dried in a Speedvac (Thermo Scientific Savant SPD121P, Asheville, NC, USA). After that, the extracts were reconstituted in 900 μL 100 % MeOH. The reconstituted samples were transferred into 2 mL Eppendorf vials with a 0.22 μm cellulose-acetate filter and centrifuged for 5 min at 16, 000× g (VWR Galaxy 16DH, Boxmeer, Netherlands). Filtrates were transferred to amber glass vials for the analysis. The LC-MS/MS analysis was performed on an Agilent 1200 LC and an Agilent 6410A QQQ (Waldbronn, Germany). The extracts were separated using a 5 μm Agilent Eclipse XDB-C18 (4.6 mm, 150 mm column, Agilent Technologies, Waldbronn, Germany) at 40 °C. The mobile phase consisted of Millipore water (v/v, eluent A) and acetonitrile (v/v, eluent B) both containing 0.1 % formic acid at a flow rate of 0.5 mL/min with the following gradient program: 0–2 min 30 % B, 6–12 min 90 % B, with a linear increase of B between 2 and 6 min and a 5 min post run at 30 % B (as described in Helsel, 2006). The injection volume was 10 µL. Identification of the eight MC
46 variants (MC_dmRR, MC_RR, MC_YR, MC_dmLR, MC_LR, MC_LY, MC_LW and MC_LF, Table 1) and nodularin (NOD) was performed in the positive Multiple Reaction Monitoring (MRM) with the following transitions: MC_dmRR 512.8 m/z [M + H]+ to 135.1 quantifier, MC_RR 519.8 m/z [M + H]+ to 135.1 quantifier, MC_YR 523.3 m/z [M + H]+ to 135.1 quantifier, MC_dmLR 491.3 m/z [M + H]+ to 847.6 quantifier, MC_LR 498.3 m/z [M + H]+ to 135.1 quantifier, MC_LY 868.4 m/z [M + H]+ to 163.0 quantifier, MC_LW 891.5 m/z [M + H]+ to 163.0 quantifier, MC_LF 852.5 m/z [M + H]+ to 163.0 quantifier and NOD 825.5 m/z [M + H]+ to 135.1 quantifier (Faassen & Lürling, 2013). Mass spectrometric settings are given in (Lürling & Faassen, 2013). Each MC variant was quantified against a calibration curve. The calibration curves were made using certified calibration standards obtained from DHI LAB Products (Hørsholm, Denmark). The limit of detection (LOD) and quantification (LOQ) for a 250 mL sample was: 0.0489 µg/L for MC_dmRR, 0.0358 µg/L for MC_RR, 0.0050 µg/L for MC_YR, 0.0054 µg/L for MC_dmLR, 0.0086 µg/L for MC_LR, 0.0817 µg/L for MC_LY, 0.0531 µg/L for MC_LW, 0.0206 µg/L for MC_LF and 0.0048 µg/L for NOD. Cylindrospermopsin and anatoxin analysis (laboratory) For the extraction of cylindrospermopsin (CYN, Table 1) and anatoxin-a (ATX, Table 1), 1.5 mL of 0.1% formic acid was added to the freeze-dried filters. Filters were sonicated for 10 min, shaken for 1 hour and then, centrifuged. This extraction procedure was repeated two more times and the combined supernatants were dried in a Speedvac (Eppendorf, Germany). Prior to analysis the dried extracts were re- dissolved in 1 mL 0.1% formic acid and filtered (0.2 µm, PVDF, Whatman, Maidstone, UK). LC-MS/MS analysis was carried out on an Agilent 2900 series HPLC system (Agilent Technologies, Waldbronn, Germany) coupled to a API 5500 QTrap mass spectrometer (AB Sciex, Framingham, MA, USA) equipped with a turbo-ion spray interface. The extracts were separated using a 5 mm Atlantis C18 (2.1 mm, 150 mm column, Waters, Eschborn, Germany) at 30 °C. The mobile phase consisted of water (v/v, eluent A) and methanol (v/v, eluent A) both containing 0.1 % formic acid, and was delivered as a linear gradient from 1% to 25% B within 5 min at a flow rate of 0.25 mL/min. The injection volume was 10 µL. Identification of CYN and ATX was performed in the positive MRM mode with the following transitions: CYN m/z 416.1 [M + H]+ to 194 (quantifier) and 416.1/176, and ATX m/z 166.1 [M + H]+ to 149, 166.1/131, 166.1/91, 166.1/43 (quantifier). Mass spectrometric settings are given in (Fastner et al. 2018). Certified reference standards were purchased from National Research Council (Ottawa, ON, Canada). The limit of detection (LOD) for both ATX and CYN was 0.0001 µg/L and the limit of quantification (LOQ) was 0.0004 µg/L for a 250 mL sample.
47 Code availability Custom-made codes in R 3.3.3. (2016) were used to combine the datasets, trace missing data and inconsistencies. The codes are available in Zenodo: https://zenodo.org/record/1219878#.Wtcc4S5ubRZ The GeoNode open source platform version 2.0 has been used for sharing the EMLS datasets among the partners. GeoNode is a web-based application that facilitates the visualization, download, sharing, and collaborative use of geospatial data through web services. GeoNode can be easily obtained at http://geonode.org/ as it is freely available under a GNU General Public License. QGIS 2.18 Las Palmas was used for creating, managing and uploading the ESRI shapefile layers into the GeoNode platform.
Data records The final dataset includes all lake, environmental, nutrient, pigment and toxin data in one data table. The description of each feature in the table can be found in Table 1. The data table is made freely available as a static copy, through direct download from the online Environmental Database Initiative (EDI) and it is provided under the name “EMLSdata_10Aug_afterRev_dateformated.csv” (Data Citation 1).
The data table is also available at the GeoNode platform (http://gleon.grid.unep.ch/), where it can be downloaded through the provided web services that secure accessibility to data by using interoperable standards as provided by the Open Geospatial Consortium (OGC). The OGC- compliant web services available are 1. the Web Map Service (WMS 1.1.1) and 2. the Web Map Tile Service (WMTS 1.0.0) for accessing the maps; 3. the Web Feature Service (WFS 1.1.0) for accessing vector data and 4. Catalogue Service for the Web (CSW 2.0.2) to access the metadata. These interoperable web service endpoints enable the user to easily access and/or integrate these datasets in their desktop, web-based client, or own workflows.
The user can find and download several features of the data table as vector layers under the tab “Layers”. In the interactive tab “Maps”, the user can visualise and download datasets of combined “Layers”, or create their own maps using the available layers. The vector layers are provided in several formats such as: ESRI shapefile, Geography Markup Language (GML) and Keyhole Markup Language (KML), JPEG, pdf etc.
Both the database and the GeoNode platform can be easily updated with the expected data from DNA and flowcytometry analysis (to follow). As new surveys may be organized following the protocols of this paper, this data can also be easily included in the database and GeoNode platform.
48 Technical validation All data received from field observers (i.e. data in the tables: Lake Data & Metadata, Sampled data and depth profiles) were checked by a data curator before uploaded into the database. Tracing missing data Participating data collectors provided a field datasheet and a metadata sheet for each lake. All sheets for each sampling event and lake metadata which did not have matching records were double checked, and errors were corrected when found. A custom-made code generated proofing reports for each table, highlighting which lakes or basins had missing data. Location data All latitude and longitude records for each lake were verified by checking visually the provided locations on google maps. Lakes were marked as verified under the following conditions: 1) the location on google maps, using either satellite or map view, was found for a lake which matched the provided name, 2) the location on google maps was located close to a water body (approximately 1 km or closer) with a matching name, 3) the location was in or beside a lake which had no name or a different name, but the name provided matched with names of regions in the area (i.e. this represents cases where lakes were named according to their closest city, or region), or 4) the provided location was near an unnamed lake in the correct country where there were no other lakes nearby. All other cases were considered unverified and the data collectors were contacted to provide the right location (Latitude and longitude in decimal degrees; WGS84). Laboratory Analyses The nutrient, pigment and toxin concentrations were analyzed centrally by certified laboratories that have optimized those specific analytical methods. Given that, these data are assumed to be correct.
49 Data citations 1. Mantzouki, E., Campbell, J. A., Van Loon, E. E., Visser, P. M., Konstantinou, I., Antoniou, M. G., Giuliani, G., Machado-Vieira, D., Gurjao de Oliveira, A., Spoljaric Maronic, D., Stevic, F., Zuna Pfeiffer, T., Bokan Vucelic, I., Gligora Udovic, M., Plenkovic-Moraj, A., Zutinic, P., Tsiarta, N., Blaha, L., Geris, R., Frankova, M., Christoffersen, K. S. , Warming, T. P., Feldmann, T., Kangro, K., Laas, A., Panksep, K., Tuvikene, L., Haggqvist, K., Salmi, P., Arvola, L., Fastner, J., Straile, D., Rothhaupt, K., Fonvielle, J., Grossart, H.-P., Avagianos, C., Kaloudis, T., Triantis, T., Zervou, S., Hiskia, A., Gkelis, S., Panou, M., McCarthy, V., Perello, V. C., Obertegger, U., Boscaini, A., Flaim, G., Salmaso, N., Karosiene, J., Kasperoviciene, J., Koreiviene, J., Savadova, K., Vitonyte, I., Haande, S., Skjelbred, B., Grabowska, M., Karpowicz, M., Chmura, D., Nawrocka, L., Kobos, J., Mazur-Marzec, H., Alcaraz-Parraga, P., Krzton, W., Walusiak, E., Wilk- Wozniak, E., Gagala, I., Mankiewicz-Boczek, J., Toporowska, M., Pawlik- Skowronska, B., Niedzwiecki, M., Peczula, W., Napiorkowska-Krzebietke, A., Dunalska, J., Kruk, M., Sienska, J., Szymanski, D., Budzynska, A., Goldyn, R., Kozak, A., Kwasizur, K., Messyasz, B., Pelechata, A., Pelechaty, M., Rosinska, J., Szelag-Wasielewska, E., Domek, P., Kokocinski, M., Jakubowska-Krepska, N., Madrecka, B., Kostrzewska-Szlakowska, I., Frak, M., Ochocka, A., Pasztaleniec, A., Jasser, I., Bankowska-Sobczak, A., Wasilewicz, M., Antao-Geraldes, A. M., Leira, M., Hernandez, A., Vasconcelos, V., Morais, J., Vale, M., Raposeiro, P. M., Goncalves, V., Aleksovski, B., Krstic, S., Nemova, H., Drastichova, I., Chomova, L. , Remec-Rekar, S., Elersek, T., Delgado-Martin, J., Garcia, D., Cereijo, J., Goma, J., Garcia-Murcia, A., Real, M., Romans, E., Noguero Ribes, J., Parreno Duque, D., Fernandez-Moran, E., Trapote, M. C., Vegas-Vilarrubia, T., Obrador, B., Ubeda, B., Galvez, J., Marce, R., Perez-Martinez, C., Ramos-Rodriguez, E., Cillero-Castro, C., Moreno-Ostos, E., Blanco, J. M., Rodriguez, V., Montes-Perez, J. J., Palomino, R. L., Rodriguez-Perez, E., Carballeira, R., Camacho, A., Picazo, A., Rochera, C., Santamans, A. C., Ferriol, C., Romo, S., Soria, J. M., Hansson, L.- A., Urrutia-Cordero, P., Colom-Montero, W., Mustonen, K., Pierson, D., Yang, Y., Bravo, A. G., Buck, M., Catalan, N., Verspagen, J. M., de Senerpont Domis, L., Teurlincx, S., Seelen, L., Maliaka, V., Verstijnen, Y., Demir, N., Tavsanoglu, U. N., Kocer, M. A., Celik, K., Karakaya, N., Yilmaz, M., Ozen, A., Maraslioglu, F., Fakioglu, O., Soylu, E. N., Apaydin Yagci, M., Cinar, S., Capkin, K., Yagci, A., Cesur, M., Bilgin, F., Bulut, C., Uysal, R., Koker, L., Akcaalan, R., Albay, M., Alp, M. T., Ozkan, K., Ozhan, K., Ongun Sevindik, T., Tunca, H., Onem, B., Bezirci, G., Beklioglu, M., Filiz, N., Levi, E. E., Iskin, U., Richardson, J., Lürling, M., Faassen, E. J., Karan, T., Edwards, C., Bergkemper, V., Torokne, A., Cerasino, L., Latour, D., Paerl, H. W., Carey, C. C. & Ibelings, B. W. Environmental Data Initiative https://doi.org/10.6073/pasta/dabc352040fa58284f78883fa9debe37 (2018).
50 Table 2 Column names, column description and explanation/format or unit for each feature of the lake file. Table 1. Column names, column description and explanation/ format or unit for each feature of the lake file. Column name Description Explanation or format or unit e.g. TR_BEY_I TR = country code (Turkey) Lake_ID A code given to each studied lake BEY = three first letters of lake name (Beysehir) I = basin number Date Date of sample collection YYYY‐MM‐DD LakeName Name of lake that is commonly used Name LabName The name of the laboratory that organized the local sampling Name Country Geographic location of lake Name Latitude Latitude coordinate of lake Decimal degrees (WGS84) Longitude Longitude coordinate of lake Decimal degrees (WGS84) Altitude_m Elevation of lake surface relative to sea level Metres MaximumDepth_m Maximum depth of lake Metres MeanDepth_m Mean depth of lake Metres SecchiDepth_m Secchi depth reading Metres Depth of sample collection SamplingDepth_m Metres (lower limit of the sampling interval) Metres ThermoclineDepth_m Location of the thermocline NA signifies a non‐stratified lake (no thermocline detected) SurfaceTemperature_C Lake water temperature at 0.5 m depth Degrees Celsius Averaged lake water temperature from surface until bottom of the EpilimneticTemperature_C Degrees Celsius thermocline TP_mgL Concentration of total phosphorus Milligrams per litre TN_mgL Concentration of total nitrogen Milligrams per litre NO3NO2_mgL Concentration of nitrates and nitrites (dissolved nutrients) Milligrams per litre NH4_mgL Concentration of ammonia (dissolved nutrients) Milligrams per litre PO4_ugL Concentration of ortho‐phosphate (dissolved nutrients) Micrograms per litre Chlorophylla_μgL Concentration of photosynthetic pigment chlorophyll‐a Micrograms per litre Chlorophyllb_μgL Concentration of photosynthetic pigment chlorophyll‐b Micrograms per litre Zeaxanthin_μgL Concentration of photosynthetic pigment zeaxanthin Micrograms per litre Diadinoxanthin_μgL Concentration of photosynthetic pigment diadinoxanthin Micrograms per litre Fμcoxanthin_μgL Concentration of photosynthetic pigment fucoxanthin Micrograms per litre Diatoxanthin_μgL Concentration of photosynthetic pigment diatoxanthin Micrograms per litre Alloxanthin_μgL Concentration of photosynthetic pigment alloxanthin Micrograms per litre Peridinin_μgL Concentration of photosynthetic pigment peridinin Micrograms per litre Chlorophyllc2_μgL Concentration of photosynthetic pigment chlorophyll‐c2 Micrograms per litre Echinenone_μgL Concentration of photosynthetic pigment echinenone Micrograms per litre Lutein_μgL Concentration of photosynthetic pigment lutein Micrograms per litre Violaxanthin_μgL Concentration of photosynthetic pigment violaxanthin Micrograms per litre MC_YR.μgL Concentration of cyanobacterial hepatotoxin microcystin YR Micrograms per litre MC_dmRR.μgL Concentration of cyanobacterial hepatotoxin microcystin dmRR Micrograms per litre MC_RR_μgL Concentration of cyanobacterial hepatotoxin microcystin RR Micrograms per litre MC_dmLR_μgL Concentration of cyanobacterial hepatotoxin microcystin dmLR Micrograms per litre MC_LR_μgL Concentration of cyanobacterial hepatotoxin microcystin LR Micrograms per litre MC_LY_μgL Concentration of cyanobacterial hepatotoxin microcystin LY Micrograms per litre MC_LW_μgL Concentration of cyanobacterial hepatotoxin microcystin LW Micrograms per litre MC_LF_μgL Concentration of cyanobacterial hepatotoxin microcystin LF Micrograms per litre NOD_μgL Concentration of cyanobacterial hepatotoxin nodularin Micrograms per litre CYN_μgL Concentration of cyanobacterial cytotoxin cylindrospermopsin Micrograms per litre ATX_μgL Concentration of cyanobacterial neurotoxin anatoxin‐a Micrograms per litre
51 Acknowledgements The authors acknowledge COST Action ES 1105 “CYANOCOST – Cyanobacterial blooms and toxins in water resources: Occurrence impacts and management” and COST Action ES 1201 “NETLAKE – Networking Lake Observatories in Europe” for contributing to this study through networking and knowledge sharing with European experts in the field. Evanthia Mantzouki was supported by a grant from the Swiss State Secretariat for Education, Research and Innovation (SERI) to Bas Ibelings and by supplementary funding from University of Geneva. We thank Wendy Beekman for the nutrient analysis and the University of Wageningen for covering the costs of this analysis from the personal funding of dr. Miquel Lürling. We thank Pieter Slot for assisting with the pigment analysis and the University of Amsterdam for covering the costs of the analysis through funding from the group of Prof. Jef Huisman and dr. Petra Visser (IBED). We would like to thank the Environmental Data Initiative for covering the cost of archiving the EMLS dataset. We would like to thank the Leibniz Institute of Freshwater Ecology and Inland Fisheries, Dept. of Experimental Limnology and the Aquatic Microbial Ecology Group for logistic and technical support of J. Fonvielle and H.-P. Grossart; and the Leibniz Association for financial support. Author contributions Evanthia Mantzouki coordinated the EMLS, collected data, curated the dataset, analyzed the pigment data and wrote the manuscript. Bas Ibelings conceived the idea for the EMLS, contributed to discussions throughout the study and to writing of the manuscript. James Campbell created the curation scripts, curated the dataset and wrote the manuscript. Emiel E. Van Loon and Petra Visser supervised the work of James Campbell and helped writing the manuscript. Iosif Constantinou developed the GeoNode platform, created the QGIS layers, uploaded and edited them in the platform. Maria G. Antoniou and Gregory Giuliani were involved in supervising Iosif Constantinou and brainstorming about developing the GeoNode platform. Jutta Fastner, Miquel Lürling and Elisabeth J. Faassen performed the toxin analyses. Sarah O’Leary did the graphic design and illustration of the infographic, Eilish Beirne did the graphic design of the infographic. The rest of the coauthors were responsible for data collection in lakes in their respective countries, for providing invaluable feedback and for finalizing the manuscript.
52 CHAPTER 4 A European Multi Lake Survey in one of the hottest summers on record: cooler Boreal regions develop bigger blooms than warmer Mediterranean regions
Mantzouki et al. Submitted to Glob Chang Biol
53 Abstract Global warming and eutrophication promote the formation of cyanobacterial blooms through direct and indirect effects. The relative importance of these environmental drivers in controlling blooms, however, is not fully understood. Moreover, although modelling studies hint at a potential synergistic interaction between nutrients and temperature in promoting blooms, convincing empirical data are largely lacking. For field studies, this may partly be due to the inherent challenges of sampling many lakes across large geographical regions, disallowing a direct comparison of lakes from different climatic zones. The European Multi Lake Survey (EMLS) is a grassroots initiative of 200 lake scientists from 27 European countries. The EMLS used standardized sampling protocols and centralized laboratory analyses to collect data on cyanobacteria and their eukaryotic competitors in the phytoplankton from lakes of different climatic zones, trophic status and depth. Our analyses reveal that in the unusually hot summer of 2015, the cooler Boreal regions exhibited significantly higher algal and cyanobacterial concentrations than the warm Mediterranean regions at comparable nutrient availability. Furthermore, the underwater light climate, an indirect consequence of changes in nutrients, significantly explained algal biomass concentrations in shallow lakes. In deep stratified lakes, the light climate interacted synergistically with water column stability, as an indirect effect of temperature, in explaining variation in phytoplankton biomass. This study demonstrates that although eutrophication and climate change are established drivers of blooms, the variety of mechanisms involved, as well as interactions between them, highlights the complex nature of cyanobacterial bloom development. We demonstrate that a well- designed grassroots effort can provide unique information at a continental scale on lake ecosystems that promotes a deeper understanding of the drivers of blooms, which ultimately is the basis for effective management of future lake ecosystems with increasing cyanobacterial blooms.
Keywords: algal blooms, cyanobacteria, biogeography, nutrients, temperature, lake stratification, photosynthetic pigments, indirect effects, synergistic interaction, grassroots science
54 Introduction Lakes are excellent sentinels of environmental change, as they integrate information from the catchment (Adrian et al., 2009; Williamson, Dodds, Kratz, & Palmer, 2008). Physical, chemical and biological properties of lakes respond to environmental perturbations such as eutrophication and climate change (Williamson et al., 2014). The main lake response to these drivers is the excessive growth of phytoplankton taxa such as cyanobacteria. Cyanobacterial blooms are expected to persist or even increase in a warmer future (Paerl, Hall, & Calandrino, 2011), the so-called “Cyanobacteria- like-it-hot” paradigm (Elliott, 2012; Kosten et al., 2012; Paerl & Huisman, 2008). Understanding cyanobacterial dynamics is of high importance, as cyanobacteria can dominate lakes and cause problems of ecological and societal concern by reducing biodiversity, promoting anoxic conditions and producing harmful toxins. Although there are many proposed mechanisms through which nutrients and temperature may promote cyanobacteria (Huisman et al. 2018), the relative importance of these drivers under various conditions is still not fully resolved. Climate warming and eutrophication may interact synergistically - i.e. the combined effect of the two stressors will be greater than the sum of their separate effects (Folt, Chen, Moore, & Burnaford, 1999) - and thus intensify cyanobacterial blooms (Moss et al., 2011). Modelling and experimental studies clearly hint at such a possible synergistic interaction between increased nutrients and temperature in promoting cyanobacterial blooms (Rigosi, Carey, Ibelings, & Brookes, 2014) and their toxin production (Lürling, van Oosterhout, & Faassen, 2017). Yet, convincing field data that support such interaction are limited to very few studies, possibly due to the inherent challenges of sampling many lakes at the same time. This impedes the direct comparison of lakes across contrasting climatic zones and lake types, which could provide better insight into the drivers of bloom development and the strength and mode of nutrient-temperature interactions. Eutrophication and climate change may promote cyanobacterial development through direct and indirect effects. Direct effects mainly impact the physiological properties of organisms, i.e. enhancing growth through nutrient availability and warmer temperatures, while indirect effects modify the physical environment that can facilitate growth. Eutrophication, apart from supporting growth, leads to a deterioration of the underwater light climate. This loss of light, as an indirect effect of nutrients, is considered the major driver of regime shifts towards cyanobacterial dominance in shallow lakes (Scheffer, Rinaldi, Gragnani, Mur, & van Nes, 1997). Climate warming can support cyanobacterial blooms either through higher water temperatures, or through increases in nutrient loading linked to higher precipitation (De Senerpont Domis et al., 2013) or through longer water residence times due to drought (Beklioglu & Tan, 2008; Reynolds, 2006), based on the location of the lake. In deep stratified lakes, climate warming enhances water column stability (Gerten & Adrian, 2002; D. M. Livingstone, 2003), which changes important biological properties of the system such as the composition (Posch, Köster, Salcher, & Pernthaler, 2012;
55 Salmaso et al., 2014) and diversity (Pomati, Matthews, Jokela, Schildknecht, & Ibelings, 2012; Straile, Jochimsen, & Kümmerlin, 2013) of the plankton. Cyanobacteria are a group of strong competitors within the phytoplankton community, possessing unique functional traits that allow them to dominate in highly productive lakes (Carey, Ibelings, Hoffmann, Hamilton, & Brookes, 2012; Mantzouki, Visser, Bormans, & Ibelings, 2016). Functional traits, such as efficient light harvesting, nitrogen fixation, polyphosphate storage and buoyancy regulation, provide tolerance to stressful environmental conditions characterised by resource deficiency (Huisman et al., 2005; Reynolds, Huszar, Kruk, Naselli-Flores, & Melo, 2002). Cyanobacteria also benefit from warmer waters (Paerl, 2014; Visser, Ibelings, Bormans, & Huisman, 2016). The indirect effects of warming, like enhanced water column stability, may be even more significant than the direct effects in promoting cyanobacteria (Carey et al., 2012). The formation of cyanobacterial blooms in freshwater systems is therefore the outcome of a complex interplay between regional (climate) and local settings (nutrients, lake morphometry, phytoplankton community composition, grazers) and cyanobacterial functional traits (e.g. buoyancy, toxin production). To disentangle the importance of these factors, multi-lake surveys are a powerful tool to explore realistic environmental gradients across wide geographic scales by following a space-for-time substitution (Beaulieu, Pick, & Gregory-Eaves, 2013; Rigosi et al., 2014; Weyhenmeyer & Jeppesen, 2009). In this approach, contemporary spatial phenomena are studied as an alternative to unavailable long-term biological records that are necessary for robust temporal analysis (Pickett, 1989). In snapshot multi-lake surveys, the study systems are sampled only once during a predefined time-period, to avoid confounding effects of seasonality. As such, we can obtain a valuable synoptic picture of the environmental factors and related biological response patterns that shape the phytoplankton communities at the continental scale. The statistical power that is generated by sampling many lakes can overcome the risk of idiosyncratic results that are lake-specific. Although studies have shown that space-for-time substitution can explain certain temporal patterns (Blois, Williams, Fitzpatrick, Jackson, & Ferrier, 2013), complementary approaches are necessary to understand the complex cause- effect relationships that induce harmful cyanobacterial blooms and predict future states. In our study, we harnessed the power of grassroots, networked science, bringing together nearly 200 scientists from 27 different countries for a coordinated, highly standardised European multi-lake survey (EMLS). We exploit the extensive EMLS dataset to disentangle the importance of various drivers of cyanobacterial blooms. Our overarching goal was to distinguish direct from indirect effects of eutrophication and climate change on cyanobacteria and competing eukaryotic algae, and to understand if and how the main drivers – nutrients and temperature - interact, i.e. what might be the dominant mode of interaction. We set out to test the following hypotheses:
56 (1) Both direct and indirect effects of eutrophication and climate change determine phytoplankton community structure at the continental scale.
(2) Synergistic interactions between temperature and nutrient effects contribute significantly to driving cyanobacterial and algal biomass.
(3) Drivers and/or their interactions differ between climatic zones and lake depth types.
Materials & Methods EMLS organisation The European Multi Lake Survey (EMLS) was carried out in summer 2015. We snapshot-sampled 369 lakes for various chemical, physical and biological parameters, thereby covering the major geographical and climatic regions in Europe (Figure 1). We used highly standardized sampling protocols and worked with dedicated laboratories that centrally analysed all the key parameters, ensuring data comparability and a fully integrated dataset. For detailed sampling methods and analyses, see Mantzouki et al. accepted in Scientific Data. In summary, the lake sampling point was defined as either the historical sampling point for which long-term records exist or the central point of the lake, if the lake had no monitoring history, or close to the dam if it was a reservoir. The sampling period was defined as the warmest two-week period of the summer, based on long-term (minimum 10 years) air temperature data in each region. An in- situ temperature profile determined the location of the thermocline and thus the sampling depth. The thermocline was defined as the point where there was ≥ 1 °C change of temperature per meter lake depth. An integrated water sample, which will henceforth be referred to as epilimnetic sample, was obtained from 0.5 m depth to the bottom of the thermocline using a water sampler. In non-stratified shallow lakes, an integrated sample was drawn from 0.5 m below the lake surface to 0.5 m above the lake bottom. As expected with such a large sampling effort, samples for certain analyses (e.g. nutrients or pigments) were not available for all waterbodies, forcing us to limit our statistical analyses to 169 EMLS lakes (Figure 1). This way we built concise statistical model for all our response variables, and avoided missing data in hypothesis testing.
57
Figure 3. Location of 169 EMLS lakes distributed over the main climatic zones of the European continent (IPCC predicted shifts 2000-2025, (Rubel & Kottek, 2010)). The Mediterranean region (n=40) consists of Csa and Csb classes (C-warm temperate, s- summer dry, a- hot summer, b- warm summer), the Continental region (n= 89) of Cfa and Cfb (f-fully humid, rest as above) and the Boreal region (n=39) of Dfb and Dfc (D-snow, c- cool summer, rest as above). Only the climatic zones with EMLS lakes are coloured.
Nutrient analysis Total phosphorus and nitrogen were measured in unfiltered samples. Sample bottles were acid washed overnight in 1M HCl and rinsed with demineralized water before usage. Nutrients were measured using a Skalar SAN+ segmented flow analyser (Skalar Analytical BV, Breda, NL) with UV/persulfate digestion integrated in the system. The total phosphate was analysed conforming to (NEN, 1986) and the total nitrogen according to (NEN, 1990). All nutrient analyses were performed at the University of Wageningen, Netherlands. Pigment analysis The analysis of pigments was modified from the method described by (Van der Staay, Brouwer, Baard, & van Mourik, 1992). All filters were freeze dried for 6 hours. Filters (45 mm GF/C and GF/F) were cut in half, placed in separate Eppendorf tubes, and kept on ice until the end of the procedure. 600 µl of 90% acetone were added to each tube along with a small amount of 0.5 mm beads. To release the pigments from the
58 phytoplankton cells, filters were placed on a bead-beater for one minute. Next, they were placed in an ultrasonic bath for ten minutes to increase the extraction yields. This procedure was repeated twice to ensure a complete extraction of the total pigment content of the filters. To achieve binding of the pigments during the High- Performance Liquid Chromatography (HPLC) analysis, 300 µl of a Tributyl Ammonium Acetate (1.5%) and Ammonium Acetate (7.7%) mix were added to each tube. Lastly, samples were centrifuged at 15,000 rpm and 4°C for 10 minutes. 35 µl of the supernatant from both Eppendorf tubes were transferred into a HPLC glass vials. Pigments were separated on a Thermo Scientific ODS Hypersil column (250 mm × 3 mm, particle size 5 μm) in a Shimadzu HPLC and using a KONTRON SPD-M2OA diode array detector. The different pigments were identified based on their retention time and absorption spectrum and quantified by means of pigment standards. Pigment analysis was performed at the University of Amsterdam, Netherlands. Statistical analyses Response variables and Environmental Predictors. Our focal response variables were chlorophyll-a as a proxy for total phytoplankton biomass, chlorophyll-b as a proxy for total green algal biomass and zeaxanthin as a proxy for cyanobacterial biomass (Figure S1, Table S1). As direct environmental predictors, we used total phosphorus (TP), total nitrogen (TN), epilimnetic temperature (TEpi) and the interactions of TP and TN with TEpi (TP * TEpi and TN * TEpi, respectively). We used maximum buoyancy frequency (BuoyFreq) and light climate (Zeu/Zmix) as indirect effect of temperature and nutrients, respectively. We also tested their interaction (Zeu/Zmix * BuoyFreq). Table 3 provides a list of all tested response variables and environmental predictors. We calculated: (1) maximum buoyancy frequency (BuoyFreq), which is a metric of stratification strength (Leach et al., 2017) using the rLakeAnalyzer R package (Winslow et al., 2016). Buoyancy frequency is defined as N2 = - (g/ρ0) x (∂ρ/∂z), where g is the gravitational acceleration, ρ0 is the density at each depth, and ∂ρ/∂z is the density gradient. In the rLakeAnalyzer, temperature profiles are used to estimate profiles of buoyancy frequency.
(2) the ratio (Zeu/Zmix) of euphotic depth (Zeu) over mixing depth (Zmix), which describes the light climate that phytoplankton experience while circulating underwater (Scheffer et al., 1997). We calculated Zeu = 2 x ZSD (Secchi depth), an average estimate from the range of constants reported in literature (e.g. Brentrup et al., 2016; Koenings & Edmundson, 1991; Salmaso, 2002). In stratified lakes, Zmix was the top of the thermocline, defined as the depth where the steepest density gradient was found (Winslow et al., 2016). In non-stratified shallow lakes, we used the sampling depth as Zmix.
59 Lake groups. We defined lake groups based on: (i) climatic zone (Figure 1) using the Köppen-Geiger’s classification (Köppen, 1900). This classification regards the main climate of the region (e.g. C=warm temperate, D = alpine), precipitation levels (e.g f = fully humid, s = summer dry) and mean temperature (a = hot summers, b = warm summers). For easier interpretation and more statistical power, we combined climatic regions where the main climate and precipitation were the same, to three main ones: Mediterranean (Csa and Csb, n=40 lakes), Continental (Cfa and Cfb, n=89 lakes) and Boreal (Dfb and Dfc, n=39 lakes) (Figure 1). In each group, the mean temperature conditions differed allowing us to test a temperature gradient. In Figure 1, we used world maps covering the period of 2000-2025 showing a shift in climatic zones based on different IPCC scenarios (Rubel & Kottek, 2010). (ii) depth type. We categorised the EMLS lakes into shallow (<6 m maximum depth, n=74 lakes) and deep (>6 m maximum depth, n=95 lakes). This classification was used in previous snapshot surveys as an approximation for weakly or strongly thermally- stratified systems, respectively (Beaulieu et al., 2013; Kosten et al., 2012).
Differences in mean values of TEpi, Zeu/Zmix, TN, TP, chlorophyll-a, chlorophyll-b and zeaxanthin within climatic zones and depth types were tested using one-way analysis of variance (ANOVA). Variables were log-transformed to follow normal distribution. Homogeneity of variance was tested using the Levene’s test from the car R package (v1.2-13). In case of heterogeneity, a Kruskal Wallis test was used instead of ANOVA. Post-hoc pairwise comparisons, using Tukey HSD (honest significant difference) for unequal sample sizes, were performed. A Games-Howell post-hoc test was used (userfriendlyscience R package) when variance was heterogeneous. Multiple linear regression model.
We log-transformed all predictors (except for TEpi) and response variables, to achieve a normal and homogeneous distribution. We used the multiple linear regression model:
Y = A0 + A1 XTP + A2 XTN + A3 X TEpi + A4 XBuoyFreq + A5 XZeu/Zmix + A6 XTP * X TEpi + A7 X TN* XTEpi + A8 XBuoyFreq * XZeu/Zmix + ε, (1) where Y represents one of the response variables (chlorophyll-a, chlorophyll-b and zeaxanthin), A0 the intercept term, A1-A8 are model parameters for each respective predictor in the models, “*” denotes the interaction between two terms, and ε is an error term. We used stepwise selection to arrive at a final model, comparing AIC scores using a modified equation that corrects for sample size and is less sensitive to outliers (R code provided by Statoo Consulting, Switzerland). If the interaction term was significant (p = 0.05), we included the lower order terms in the equation. The most parsimonious model, in which elimination or addition of any other predictors would not improve the model by ΔAIC > 2, was used for the analysis of variance. Four to five
60 other models with a ΔAIC < 2 from the most parsimonious model were also tested, to ensure that significant predictors were always the same (data not shown). The metric “lmg” of the relaimpo R package (Groemping, 2006) was used to decompose the overall R2 of each final model, into the absolute contributions of each predictor term and their interaction terms (similarly done in (Martín, Doering, & Robinson, 2017; Rigosi et al., 2014). The relative contribution of its predictor was normalized and summed to the total R2. A bootstrapping approach was used to resample the observed data 9999 times to determine if there were any differences among the 95% confidence intervals of each of the predictors in the interaction terms. If those differences did not include zero, it indicated that the predictors were significantly different from each other. When the interaction term had a significant value of p < 0.05 and was positive, this was interpreted as an indicator of a synergistic interaction. To avoid multicollinearity between the interactions and their main effects, we checked the Variance Inflation Factor (VIF). If they were exhibiting high numbers (VIF>3), we centred the interaction term with the mean of the raw variables which alleviated the collinearity problem.
Results The EMLS lakes exhibited a wide variation in physical, chemical and biological parameters (Table 1). Epilimnetic temperature (TEpi) ranged from 9 to 32 °C with a median value of 21 °C. The median lake had a eutrophic status (60 μg TP/L). The light climate, as a ratio of Zeu/Zmix, ranged from 0.02 to 10.8 with a median of 0.63. The maximum buoyancy frequency, as a proxy of lake water stability, ranged from 0.00003 to 0.03 s-2, with a median value of 0.004 s-2. This shows a considerable level of lake stratification for half of the studied lakes. There was no significant correlation among the environmental predictors (Fig S2), which allowed us to use all the environmental predictors together in our models. The response variables of the EMLS lakes also varied substantially. Chlorophyll-a ranged over three orders of magnitude (0.6-690 μg/L; Table 1) with a median value of 13.6 μg/L. According to WHO, 50 % of the EMLS lakes fall within moderate ecological status, which could be of a potential risk if 50 % of the phytoplankton biomass is cyanobacteria. Zeaxanthin and chlorophyll-b ranged over two orders of magnitude (0.07-53.52 and 0.00-111.78 μg/L, respectively) with a median value of 1 μg/L for both pigments. Figure 1 shows the distribution of the EMLS lakes over the European continent. A total of 40 Mediterranean, 89 Continental and 39 Boreal lakes were included in the subsequent analysis. Analysis of variance on a selection of climatic zones, has a clear advantage over a simple latitudinal analysis, as several lakes within the Continental region are classified as Boreal lakes, based on their climatic characteristics, rather than their position on a latitudinal gradient.
61 Environmental predictors by climatic zone and depth type. The Mediterranean region exhibited significantly higher mean epilimnetic temperatures (TEpi) than the Continental and Boreal (F2,166 = 27.48, p < 0.0001) lakes (Figure 2 a) by 3.8 and 4.5 °C, respectively. There was no significant difference between the Continental and Boreal epilimnetic temperatures (p = 0.67). Mean Zeu/Zmix (F2,166 = 1.865, p = 0.158), TN (F2,166 = 2.042, p = 0.133) and TP (F2,166 = 0.515, p = 0.599) did not vary significantly among the climatic zones (Figure 2b, c and d). Mean TEpi was 2.3 °C (F1,168 = 14.29, p = 0.0002) higher in shallow lakes compared to deep lakes (Figure 2 e). Similarly, shallow lakes had a significantly better light climate (Zeu/Zmix ratio) compared to deep lakes (F1,168 = 6.912, p = 0.0094) (Figure 2 f). There was no significant difference in mean TN or TP between the two different depth types (F1,168 = 3.523, p = 0.0623, Figure2 g; F1,168 = 1.189, p = 0.277, Figure2 h, respectively). Table 3. A list of all the variables included in the analyses, as well as their units, range of values, means, medians and standard deviations.
Variable N lakes Units Range Mean Media StdD n
Maximum Depth 163 m 0.5-310.00 19.97 8.00 39.53 Total Nitrogen 169 mg/L 0.07-5.53 1.00 0.77 0.84 Total Phosphorus 169 mg/L 0.02-0.87 0.11 0.06 0.13 Epilimnetic 169 °C 8.67-31.23 21.29 20.87 3.78 Temperature Secchi Depth 169 m 0.12-7.96 1.53 1.03 1.50
Zeu/Zmix 169 - 0.02-10.83 1.03 0.63 1.33 Max. Buoyancy 169 s-2 2.90 E-05-0.03 0.005 0.004 0.004 Frequency Chlorophyll-a 169 μg/L 0.61-690.16 43.10 13.60 82.75 Zeaxanthin 169 μg/L 0.07-53.52 3.34 1.00 6.22 Chlorophyll-b 1691 μg/L 0.00-111.78 4.03 1.03 11.60
1Chlorophyll-b was absent in 27 lakes (0.00 μg/L). These lakes were excluded in the chlorophyll- b models to ensure normality.
62
Figure 2. Boxplots of predictors variation across the different climatic zones: Med- Mediterranean (n = 40 lakes), Cont- Continental (n = 89) and Bor- Boreal (n = 39), and the different depth types- shallow (n = 74) and deep (n = 95). a, e: epilimnetic temperature (TEpi); b, f: log light climate (Zeu/Zmix); c, g: log total nitrogen (TN); and d, h: log total phosphorus (TP). The p values refer to the one-way ANOVA between the climatic zones or depth types, significant results are marked with red. Different symbols in blue indicate significant differences among categories (Tukey test; p < 0.05).
63 Response variables by climatic zone and depth type. Both total chlorophyll-a (proxy for total phytoplankton biomass) and zeaxanthin (proxy for cyanobacterial biomass) concentrations differed among regions increased significantly from the southern Mediterranean regions to the northern Boreal regions (F2,166 = 11.22, p < 0.0001; F2,166 = 3.68 p = 0.027; Figure 3a and c, respectively) by 1.45 μg/L for chlorophyll-a and 0.72 μg/L for zeaxanthin. From the Continental to the Boreal regions only chlorophyll-a increased significantly (p = 0.002) by 0.94 μg/L. There were no significant differences in the mean concentration of chlorophyll-a and zeaxanthin between Mediterranean and Continental lakes (p = 0.10; p = 0.68, respectively). Mean chlorophyll-b (proxy for green algal biomass) did not exhibit any significant changes in its mean values across the different climatic zones (F2,166 = 1.56, p = 0.2). In shallow lakes, chlorophyll-a, chlorophyll-b and zeaxanthin (Figure 3d, e and f, respectively) had significantly higher mean concentrations by 1.15, 1.30 and 1.22 μg/L, respectively, compared to deep lakes (F1,168 = 27.54, F1,168 = 33.71, F1,168 = 61.26 and p < 0.0001, respectively).
64
Figure 3. Boxplots of variation in response variables across the different climatic zones: Med- Mediterranean (n = 40 lakes), Cont- Continental (n = 89) and Bor- Boreal (n = 39), and the different depth types- shallow (n = 74) and deep (n = 95). a, d.: log chlorophyll-a; b, e: log chlorophyll-b; c, f: log zeaxanthin. The p values refer to the one-way ANOVA between the climatic zones or depth types, significant results are marked with red. Different symbols in blue indicate significant differences among categories (Tukey test; p < 0.05).
65 EMLS multivariate analysis of environmental predictors. In the overall dataset, the importance of different environmental drivers varied for chlorophyll-a, chlorophyll-b and zeaxanthin. Whereas total nitrogen, light climate and maximum buoyancy frequency explained a significant proportion of the overall R2 for chlorophyll-a (Table 2a), epilimnetic temperature and maximum buoyancy frequency were the most important determinants of chlorophyll-b and zeaxanthin (Figure4a). We found a significant synergistic interaction between light climate and buoyancy frequency for chlorophyll-a and zeaxanthin (Figure4a), but not for chlorophyll-b. A much smaller but significant proportion of the overall R2 for chlorophyll-a and zeaxanthin was explained by total phosphorous. The contribution of environmental drivers to the phytoplankton biomass differed between climatic zones. In Mediterranean lakes, while epilimnetic temperature, maximum buoyancy frequency and total phosphorous explained most variation of chlorophyll-a (Table 2b), total nitrogen and its interaction with epilimnetic temperature explained most variation in chlorophyll-b (Figure 4b, Table 2b). Variation in zeaxanthin was determined mostly by temperature and total phosphorus, in Mediterranean lakes (Table 2b, Figure 4b). In the Continental climatic zone, epilimnetic temperature and maximum buoyancy frequency explained the highest percentage of variation in tested pigment concentrations (Table 2b). Specifically, temperature was the only predictor that positively determined variation in zeaxanthin among lakes in the Continental zone (Figure 4c, Table 2b). Stratification strength, expressed as maximum buoyancy frequency, significantly determined variation in chlorophyll-a and -b (Figure 4c). In Boreal lakes, epilimnetic temperature and maximum buoyancy frequency were the most significant predictors for chlorophyll-b (Figure 4d) while they contributed significantly to the overall R2 in the chlorophyll-a model (Table 2b). Maximum buoyancy frequency in interaction with light climate were the most significant predictors for chlorophyll-a (Table 2b) while it was the only significant predictor for zeaxanthin, in the boreal zone (Figure 4d). Other drivers such as total nitrogen and its interaction with epilimnetic temperature contributed less but still significantly to the overall R2 in the chlorophyll-a and -b Boreal models (Figure 4d; Table 2b).
Shallow lakes responded strongly to the local light climate, with the predictor Zeu/Zmix explaining large proportion of the overall model R2 for chlorophyll-a, chlorophyll-b and zeaxanthin (Figure 4e, Table 2c). Variation in algal and cyanobacterial biomass was also significantly determined by total phosphorus, in shallow lakes (Figure 4e, Table 2c). Total nitrogen explained the largest proportion of the overall model R2 for green algal biomass (Figure 4e, Table 2c) in shallow lakes. The predominance of a positive interaction between light climate and maximum buoyancy frequency was not pigment specific in deep lakes. This positive interaction was significant for chlorophyll-a, chlorophyll-b and zeaxanthin (Figure 4f, Table 2c). Light climate also
66 had a significant effect on variation in chlorophyll-a, -b and zeaxanthin biomass, in deep lakes (Figure 4f, Table 2c). Table 2. Summary of regression models showing the effect of total nitrogen (TN), total phosphorous (TP), epilimnetic temperature (TEpi), maximum buoyancy frequency (BuoyFreq), light climate (Zeu/Zmix) and the interactions TN * TEpi, TP * TEpi and Zeu/Zmix * BuoyFreq on chlorophyll-a, chlorophyll-b and zeaxanthin in (a) the overall dataset, (b) different climatic zones (Mediterranean, Continental, Boreal) and (c) different depth types (Shallow, Deep). We provide the overall R2 of the model, and the decomposed R2 and significance, p, for each predictor. For the interactions, we include the 95% confidence interval differences between the two terms. Significant results are marked in red and bold. Statistical details of the linear models are provided in Table S2.
67
68 Figure 4. Bar charts of the R2 contribution (R2 partitioned) of the environmental predictors: total phosphorus (TP), total nitrogen (TN), epilimnetic temperature (TEpi), underwater light climate (Zeu/Zmix), maximum buoyancy frequency (BuoyFreq), and interactions of TP * TEpi, TN * TEpi and Zeu/Zmix * BuoyFreq to the overall R2 of the multivariate models. The relative importance model identified which predictors explain most variance of the three representative pigments (Chlorophyll-a, Chlorophyll-b and Zeaxanthin) for: a) overall – all lakes together, b) Mediterranean, c) Continental, d) Boreal climatic zones and e) Shallow, f) Deep depth types. Only the significant predictors that contributed the highest to the overall R2 are shown. More information is provided in Table 2.
69 Discussion In the EMLS, we sampled lakes throughout the European continent in a collaborative effort to study the relative roles of two key environmental stressors – eutrophication and global warming - on lake phytoplankton, in particular the occurrence of cyanobacterial blooms. In this discussion, we focus on the key results that address our hypotheses. Clearly, in a dataset as large as this, there are other potentially interesting results and relationships that we do not address in this paper. As the complete dataset will be publicly available (Mantzouki et al. in press Scientific Data), we encourage other researchers to further explore the EMLS dataset. Direct and indirect effects of nutrients. We hypothesized that variation in abundance and composition of lake phytoplankton at the continental scale is the outcome of direct and indirect drivers that are linked to eutrophication and climate warming. Based on previous multi-lake surveys we expected that nutrients would explain more variance in chlorophyll-a, -b and zeaxanthin than temperature (Rigosi et al., 2014, Beaulieu et al., 2013), but this was not observed for our EMLS data. Analysis showed that total nitrogen, rather than total phosphorus, explains about 15 % of the overall R2 for chlorophyll-a and -b. Total nitrogen being a better predictor of algal biomass is in agreement with previous studies (Dolman et al., 2012; Downing, Watson, & McCauley, 2001). In the NLA dataset, total nitrogen over total phosphorus, better explained biovolume in certain taxa (Rigosi et al., 2014). Previous studies found that high biomass of nitrogen fixing cyanobacteria tend to have high N: P ratio, explaining TN as chlorophyll-a predictor (Klausmeier, Litchman, Daufresne, & Levin, 2004). On the other hand, Beaulieu et al., (2013) showed that for predicting chlorophyll-a, total nitrogen and phosphorus were equally important. Our results support most studies, concluding that nitrogen must be taken into account when managing cyanobacterial blooms (Conley et al., 2009). DNA analyses of the phytoplankton communities in the EMLS samples could elaborate further.
We found that the light climate (ratio Zeu/Zmix) determined a high percentage of the overall R2 in the chlorophyll-a model (Figure 4a). Light climate, in particular in productive shallow lakes can be seen as a more integrative parameter of nutrient loading than a snapshot sample of total nutrients, while Zmix gives an indication of the recent mixing history of the lake. In (hyper)eutrophic lakes, light often becomes the limiting resource and is an important factor in favouring cyanobacteria (Huisman et al., 2004). The 84 % of the EMLS lakes were eutrophic and hypereutrophic. Individual countries contributed data from lakes that they routinely sample, which tend to have a history of eutrophication issues. The bias towards productive lakes is a potential disadvantage of the grassroots EMLS, compared to a centrally organised survey like the NLA, where a Generalized Random Tessellation Stratified Survey Design (Stevens Jr & Olsen, 2004) was used. However, it also reflects the reality where eutrophic lakes still are generally more common in Europe. Understanding bloom dynamics in
70 eutrophic systems is therefore arguably more relevant for water management due to predominance of these systems. Direct and indirect effects of temperature. The multiple regression models showed a highly significant contribution of both the direct and indirect effects of temperature in explaining cyanobacterial (zeaxanthin) and green algal (chlorophyll-b) biomass. Epilimnetic temperature (direct effect) explained 28 and 32 %, while maximum buoyancy frequency (indirect effect) explained 25 and 33 % of the overall R2 for zeaxanthin and chlorophyll-b, respectively (Figure 4a, Table 2a). At the time of sampling, the summer of 2015 was the third warmest summer (after 2003 and 2010) since 1880 in Europe (NOAA). In general, the European continent showed a + 1.57 °C surface temperature anomaly during the entire summer. This unusually prolonged warm period in the European continent might change the relative roles of nutrients and temperature in explaining variation in phytoplankton pigments in the 2015 EMLS dataset compared to years with more average meteorological conditions. Maximum buoyancy frequency has been shown to be a strong predictor of deep chlorophyll maxima in stratified lakes (Leach et al., 2017), but this also seems to hold true for a dataset where shallow lakes (42 % of the tested lakes) are included. Under conditions of micro-stratification, buoyant cyanobacteria such as Microcystis, manage to position themselves in the illuminated near-surface mixed layer, increasing their access to light, compared to non-buoyant competitors (Ibelings, Mur, Kinsman, & Walsby, 1991). In a future climate, the strength of micro-stratification is expected to increase, and since it is predicted that night-time warming will outpace daytime warming (Davy & Esau, 2016), the breakdown at night through convectional cooling will diminish, leading to prolonged periods of water column stability in shallow lakes, further promoting buoyant cyanobacteria. Synergistic interactions. We hypothesized that temperatures and nutrients may interact synergistically in promoting blooms. The processes involved in this interaction can be linked to both the direct and indirect effects of the two main drivers. In the overall dataset, although total phosphorus in interaction with epilimnetic temperature was retained in the models of chlorophyll-a and zeaxanthin (Table 2a), this interaction was not a significant predictor. Significant synergistic interactions between total nutrients and epilimnetic temperatures were found in Mediterranean and Boreal lakes (see discussion further down). Interestingly, we found strong support for synergistic interactions between the indirect effects of temperature and nutrients. The interaction between light climate (Zeu/Zmix) and water column stability (BuoyFreq) had the highest contribution to the overall R2 for chlorophyll-a (Table 2a). A lower but significant contribution of the interaction between light and maximum buoyancy frequency was found for
71 zeaxanthin (Figure 4a). High algal biomass increases turbidity, which subsequently can increase water temperature locally through increased heat absorption (Ibelings, Vonk, Los, Van der Molen, & Mood, 2003). This positive feedback can reinforce stratification (Paerl & Huisman, 2008) offering a potential explanation for how light climate can potentially interact synergistically with stratification (Figure 4a, d, f). Environmental drivers in different climatic zones. The one-way analysis of variance of the climatic zones showed that there was no significant variation in nutrients among the zones (Figure 2a, c and d); therefore, continental scale variation should mainly be driven by epilimnetic temperature differences. Epilimnetic temperature emerged as a dominant factor explaining variation in abundance of green algae and cyanobacteria (Figure 4b, c, d). We found a significant variation in the epilimnetic temperature of lakes across Europe, with lakes in the Mediterranean region, not surprisingly, being significantly warmer than the Continental and the Boreal lakes (Figure 2a). Both field monitoring and experimental studies have previously shown that cyanobacteria proliferate when temperatures increase over a latitudinal gradient, promoting the idea that blooms will expand in response to global warming (Elliott, 2012; Kosten et al., 2012; Paerl & Huisman, 2008). Yet, surprisingly, our data show that cyanobacterial biomass was higher in cooler Boreal regions than in the warmer Mediterranean region under comparable nutrient availability (Figure 3a and c). In the Continental region, epilimnetic temperature determined most of the variance in total chlorophyll-a, chlorophyll-b and zeaxanthin (Figure 4b). Maximum buoyancy frequency, as an indirect effect of temperature, also had a significant negative effect on chlorophyll-a and -b (Table 2b, Table S2). Previous multi-lake surveys showed that temperature significantly affected cyanobacterial biomass independently of other environmental predictors (Beaulieu et al., 2012, Rigosi et al., 2014). Temperature in the warmer Mediterranean lakes, however, did not promote higher algal or cyanobacterial biomass (Figure 3a, b and c). The distinct difference in the role of temperature between the more southern and northern regions in Europe can possibly be explained by regional variation in temperature anomalies during the period of sampling in the EMLS. Boreal regions exhibited a temperature anomaly of +5 °C against the long-term average, from the 2nd until the 24th of August, during the period where 77 % of lakes in these regions were sampled. In contrast, in the Mediterranean and Continental lakes the average temperature anomaly was restricted to around +2 °C. Demonstrably, the northern lakes experienced a significantly stronger temperature increase - lake temperatures closely follow atmospheric temperatures (Livingstone & Lotter, 1998) - above normal that far exceeded the European average for that summer. Cyanobacterial growth rate steeply increases with temperature until about 25 °C and levels out at about 28 °C (Paerl, 2014). Actually, the optimal growth temperature for cyanobacteria is not significantly different from eukaryotic green algae (Lürling,
72 Eshetu, Faassen, Kosten, & Huszar, 2012). However, growth rate increases more steeply with temperature in cyanobacteria compared to green algae (Visser et al., 2016). Temperature becomes damaging to cyanobacteria and algae when it exceeds ~33 °C (Paerl, 2014). This implies that an increase in lake temperature between e.g. 20- 25 °C has a larger positive effect than an increase between e.g. 25-28 °C. The exceptional temperature increase in Boreal lakes compared to a relatively “normal” situation in the Mediterranean lakes, during the peak cyanobacterial growth season, could explain why cyanobacteria in northern regions would be able to reach biomass concentrations similar to those in the southern regions, since lake water temperature came closer to the optimal temperature for growth. But why would biomass in the North exceed biomass in the South? There are strong differences in growth curves between genera and species (Huisman et al., 2018). Reaction norms even differ between strains of the same species (Gsell et al., 2012; Thomas, Kremer, & Litchman, 2016). Species evolve in response to local conditions they typically encounter, i.e. different set of reaction norms must have evolved and be maintained in the various climatic zones. The rather extreme +5 °C scenario in northern Europe may have selected strains at the extreme upper end of the reaction norms, contributing to an exceptionally positive response to the heatwaves of summer 2015. While an exceptional increase in temperature may have induced a reduction in growth rate, that obviously was not the case. This explanation for now is merely an (informed) hypothesis, but generating and subsequent testing of new hypotheses should be a key objective of large-scale field observations like EMLS. More detailed integrated lab- field studies, including both ecological and evolutionary aspects are needed to resolve this issue. Furthermore, Boreal regions showed a dominant role for the maximum buoyancy frequency interacting positively with light climate (Figure 4d). Carey et al., (2012) argued that indirect effects of warming would be more important than direct effects. It is possible that the prolonged heat event in Boreal region lasted long enough (Hoy, Hänsel, Skalak, Ustrnul, & Bochnicek, 2016) to enhance water column stability that promoted cyanobacteria with efficient flotation behaviour. Further analysis that would reveal the predominant traits in the EMLS samples could further explain this finding. Finally, there was an indication of synergistic interactions between total nitrogen and epilimnetic temperature in the Mediterranean and Boreal lakes (Table 2b). This key result is comparable to Rigosi et al., (2014), who used data from the US National Lake Assessment (EPA NLA 2007). In the NLA dataset, the interaction between nutrients and temperature explains the largest part of the variance for the subset of eutrophic and hypereutrophic lakes. Since 75 % of the Mediterranean and 97 % of the Boreal lakes were eutrophic and hypereutrophic, we can conclude that the statistical outcome of our model was primarily driven by those lakes. Consequently, the results of this study provide additional evidence that synergistic interactions between nutrients and temperature are likely in highly nutrient-rich lakes.
73 Environmental drivers in shallow vs. deep lakes.
The underwater light climate (Zeu/Zmix) is the dominant factor that explains variation in chlorophyll-a and zeaxanthin in shallow lakes, whereas in deep lakes it interacts synergistically with maximum buoyancy frequency (Figure 4e, Table 2c). Unlike deep lakes, shallow lakes exist in clearly distinct states, clear vs. turbid. Mechanisms directly linked to the underwater light climate, e.g. high cyanobacterial biomass, and benthivorous fish stirring up the sediment, provide varying degrees of resilience to the turbid state (Scheffer et al., 1997). With 72 % of the shallow EMLS lakes having a Secchi depth less than 0.8 m, we could argue that the majority are in a turbid state, be it stable or not. This may go some way to explain the critical role of light in determining abundance of algae and especially cyanobacteria (Figure 4e). Additional top-down effects on cyanobacterial biomass linked to climate warming can provide further resilience to the turbid state. Warmer temperatures may increase the population of zooplanktivorous fish removing zooplankton that otherwise could control phytoplankton (Mooij, Janse, Domis, Hülsmann, & Ibelings, 2007). In the deep EMLS lakes, a significant interaction between the indirect effects of temperature and nutrients corroborates our third hypothesis about the importance of interactions between drivers of change. According to our results, maximum buoyancy frequency interacts synergistically with Zeu/Zmix in promoting total phytoplankton, green algal and cyanobacterial biomass (Figure 4f, Table 2c). This implies that when maximum buoyancy frequency is high, mixing is restricted to the upper layers and phytoplankton may be maintained within the euphotic zone. Hence, a high buoyancy frequency may result in good light conditions (high Zeu/Zm) for phytoplankton growth, provided that non-buoyant phytoplankton is not lost from the shallow surface mixed layer due to sedimentation (Camacho, 2006; Reynolds, 2006). In a warmer future, a stable epilimnion, although promoting access to light, could also induce nutrient limitation, as nutrients would remain locked away in the hypolimnion. This loss of nutrient availability would be strengthened by an incomplete winter overturn (Yankova, Neuenschwander, Koster, & Posch, 2017). In such cases, cyanobacteria that can regulate their buoyancy (Ganf & Oliver, 1982) and potentially enter the hypolimnion for access to nutrients, will have a clear advantage over non-migrating algal groups (Cottingham, Ewing, Greer, Carey, & Weathers, 2015; Mantzouki et al., 2016). In the EMLS, where 75 % of the deep lakes are eutrophic or hypereutrophic, nutrient limitation is not likely to be an issue, thus retaining the possibility of a synergistic interaction between buoyancy frequency and light climate (Figure 4f, Table 2c). Conversely, if buoyancy frequency is low, mixing can penetrate deeper, taking the algal communities away from the euphotic zone and restricting cyanobacterial or algal growth (Figure 5). A deeper mixed layer will allow light to reach greater depths by diluting epilimnetic phytoplankton over a larger volume of lake water, thus reducing light extinction. This deeper euphotic depth likely will not make up for light limitation
74 due to a deeper mixing depth, so the ratio Zeu/Zm would still decrease when water column stability decreases (Figure 5, Table 2c). Changes in the strength of stratification as an indirect effect of global warming constitute a crucial factor, alongside eutrophication, in defining water quality in deep lakes, today and in a warmer future (Schwefel, Gaudard, Wuest, & Bouffard, 2016). Since the expectation is indeed that water column stability will increase in the future (Gerten & Adrian, 2002) the interactions shown in Figure 5 are more likely to promote rather than decrease cyanobacterial blooms.
Figure 5. Schematic overview of how lake stratification (BuoyFreq determined as the depth where the slope of the temperature gradient is maximal) and light climate (Zeu/Zmix) define biomass in deep lakes, based on the EMLS results of the multivariate analysis (see Figure 4 as well). In panel a: increased stratification (high BuoyFreq - steep slope) limits mixing depth (Zmix), allowing phytoplankton to circulate within the euphotic depth (Zeu). In panel b: weaker stratification (low BuoyFreq – gradual slope) allows deeper mixing that has the potential to ameliorate euphotic depth through dilution of the phytoplankton concentration in comparison to panel a. However, this increase in Zeu cannot compensate for the deeper mixing, ultimately reducing the ratio of Zeu/Zmix. At this point, we would like to bring up a possible complication of the sampling strategy we followed in the EMLS, taking depth-integrated samples extending from the lake surface to bottom of the thermocline. In lakes with a deep thermocline and a phytoplankton community that is concentrated in the upper layers, taking a deep integrated sample could lead to thinning and a potential underestimation of the chlorophyll-a content of that sample. Be aware, however, that many deep lakes are actually characterized by the presence of a DCM, a Deep Chlorophyll-a Maximum (Leach et al., 2017), where taking a shallow sample would lead to serious underestimation. Considerations like this have actually lead many authors (e.g. Ptacnik, Solimini, & Brettum, 2009; Noges, Poikane, Koiv, & Noges, 2010) to use depth integrated sampling since it probably comes closest to yielding a representative
75 chlorophyll-a value of a lake from a single sample. Assuming that the risk of thinning is real it should primarily occur in the deeper EMLS lakes and in particular those with a strong density gradient. In lakes with weak density gradients and therefore fairly homogeneous chlorophyll-a distribution, the consequences of thinning by deep sampling would be minimal. Thus, given that deep lakes with a high water column stability (high BuoyFreq) are the most likely to “suffer” from thinning effects we would expect a negative correlation between BuoyFreq and chlorophyll-a, chlorophyll-b or zeaxanthin. However, model regressions show the opposite: for deep lakes, pigments correlate positively with a high BuoyFreq (e.g. Log Chla = - 0.36 + 2.50 Zeu/Zmix + 1.15 TEpi + 0.18 BuoyFreq + 0.50 Zeu/Zmix * BuoyFreq). We would tentatively interpret this positive correlation as lakes with stronger stratification being more likely to develop DCMs. When we re-ran our models including sampling depth as covariate, the same covariates as before plus sampling depth came out as significant. Hence, one would conclude that sampling depth is meaningful to explain EMLS observations. Moreover, a model that includes only sampling depth explains results as good as the models we presented. However, this should not be surprising, since some of the key ecological drivers of phytoplankton identified in our study, like the light climate, are known to strongly correlate with lake and thermocline depth, the factors that determine sampling depth for respectively shallow and deep lakes. Deeper lakes naturally tend to have lower chlorophyll-a than shallow lakes because of a deterioration in the light climate, but this reduction in pigments is the result of a meaningful driver and not an artefact of thinning. Concluding, we identify thinning in deep lakes as a possible confounding factor in multiple lake studies that include both shallow and deep lakes, but the checks we have ran, give us no reason to believe that we cannot retain the models we presented.
76 Outlook towards the future. Our results partially agree with the notion that “blooms like it hot”. In fact, according to our findings, it may be more correct to say that “blooms like it warmer” (than normal). In the overall dataset, epilimnetic temperature directly promotes cyanobacteria (Figure S3). In the climatic zones, the largest blooms were found in the cooler - but strongly warmed up - Boreal regions. Hence, temperature becomes increasingly relevant when heatwaves take water temperature well beyond the norm for that region. Since heatwave summers are expected to increase, this is a relevant observation for the development of future blooms across the continent. The significant positive interaction between total nitrogen and epilimnetic temperature in nutrient rich lakes, provides evidence for an “allied attack” between eutrophication and climate warming (Moss et al., 2011), and observational support that a synergistic interaction will be restricted to productive lakes (Brookes & Carey, 2011). Although we did not report on all possible effects of climate change, our study supports the expectation that cyanobacterial growth will increase in a warmer future. Based upon our results we retain all three hypotheses i.e. direct and indirect effects, as well as interactions between them explain the distribution of phytoplankton biomass across the continent, with different patterns emerging from the various climatic regions and lake depth type. Our data indicate that global warming and ongoing eutrophication will reinforce cyanobacterial development, more than that of green algae (e.g. green algae did not significantly increase with extreme temperature changes in the Boreal lakes), since functional traits allow them to better function under environmental change (Carey et al., 2012). Additionally, the distribution of the EMLS cyanobacterial toxins was mostly driven by direct and indirect effects of temperature (Mantzouki et al., 2018) potentially promoting more toxic species or strains in a warmer future (Dziallas & Grossart, 2011). Top down effects by increased grazing pressure on better quality food such as green algae will further facilitate cyanobacterial blooms (Ger et al., 2016). The EMLS makes a significant contribution towards a more robust model of phytoplankton response, including cyanobacteria, to drivers of change. This knowledge is needed for adaptive management of blooms at times of global change.
77 Acknowledgements The authors acknowledge COST Action ES 1105 “CYANOCOST – Cyanobacterial blooms and toxins in water resources: Occurrence impacts and management” and COST Action ES 1201 “NETLAKE – Networking Lake Observatories in Europe” for contributing to this study through networking and knowledge sharing with European experts in the field. We acknowledge the members of the Global Lake Ecological Observatory Network (GLEON) for their collaborative spirit and enthusiasm that inspired the grassroots effort of the EMLS. Evanthia Mantzouki was supported by a grant from the Swiss State Secretariat for Education, Research and Innovation to Bas Ibelings and by supplementary funding from University of Geneva. We thank Wendy Beekman for the nutrient analysis. We thank Pieter Slot for assisting with the pigment analysis. We thank the Leibniz Institute of Freshwater Ecology and the Aquatic Microbial Ecology Group for logistic and technical support of J. Fonvielle and H.-P. Grossart; and the Leibniz Association for financial support. The collection of data for Lough Erne and Lough Neagh were funded by the Department of Agriculture, Environment and Rural Affairs, Northern Ireland. Finally, we would like to thank the numerous other assistants that helped realising each local survey.
Supporting material Pigments as descriptors of phytoplankton biomass. NMDS was performed in order to evaluate the use of chlorophyll-b and zeaxanthin as proxies for total green algal and cyanobacterial biomass. We used the log- transformed total biovolumes (in μm3 ml-1) of Cyanophyceae, Chlorophyceae, Bacillariophyceae, Euglenophyceae, Dinophyceae, Conjugatophyceae, Cryptophyceae and Chrysophyceae of the available EMLS microscopical counts as response variables, and pigments were vector fitted as descriptors. Ordinations (metaMDS command) and vector fitting (envfit command) were performed in R using the vegan package 2.4-4. The number of MDS axes was chosen by minimizing stress, a measure of the mismatch of distance among response variables indicated by the Bray-Curtis matrix and distance in the ordination. We found a strong positive relationship between the taxon-specific pigments, zeaxanthin and chlorophyll-b, as descriptors of total cyanobacterial and total green algal biomass, respectively (Table S1). In Figure S1, the arrows depict the pigments and point out the phytoplankton group that was best described by the corresponding pigment. Ordination stress, minimized to 0.155 using three dimensions, showed a good match (less than 0.2) of the distance among the phytoplankton groups and the distance calculated in the ordination (distances among the groups in the graph). Zeaxanthin (p=0.0001, R2=0.24) lined up significantly with the cyanobacterial biovolume (Cyanophyceae) and chlorophyll-b (p=0.0001, R2=0.18) lined up significantly with the green algal biovolume (Chlorophyceae).
78
Figure S 1. Non-metric multidimensional scaling (NMDS) of the biovolumes of the main phytoplankton groups (in red) and the use of chlorophyll-b and zeaxanthin as predictors (in blue) of green algae (Chlorophyceae) and cyanobacteria (Cyanophyceae), respectively.
Table S 1 Results of the NMDS analysis of the pigments chlorophyll-b and zeaxanthin as predictors of the phytoplankton groups green algae (Chlorophyceae) and cyanobacteria (Cyanophyceae), respectively.
NMDS1 NMDS2 r2 Pr (>r)
Chlorophyll-b -0.99609 0.08836 0.1784 <0.0001
Zeaxanthin -0.8835 -0.46842 0.2412 <0.0001
MDS1 MDS2 MDS3
Chlorophyceae -0.3414 0.031409 0.130996
Cyanophyceae -0.60692 -0.36787 0.009844
79
Figure S 2. Correlation matrix of the predictors of chlorophyll-a, chlorophyll-b and zeaxanthin. Left panels report relationship between predictors, right panels report Spearman’s correlation coefficients and diagonal panels report distribution of the predictors epilimnetic temperature (TEpi), log total phosphorus (TP), log total nitrogen (TN), log maximum buoyancy frequency (BuoyFreq) and log light climate (Zeu/Zmix).
80 Climatic Zone Mediterranean 6 Continental Boreal
)
L
/
g
μ
(
)
n
o 4
i
t
a
r
t
n
e
c
n
o
C
a
-
l
l
y
h
p
o
r 2
o
l
h
C
(
g
o
l
0
15 20 25 30 Epilimnetic Temperature (°C)
Figure S 3 Scatterplot of the relationship between epilimnetic temperature – TEpi (°C) and log chlorophyll-a concentration (μg/L) in the 169 EMLS lakes. The different climatic zones are colour marked (green – Mediterranean, red – Continental and blue – Boreal). Linear regression: logChla= +1.45 +0.06 ET, p = 0.07, AdjR2 = 0.013
81 Table S 2. Predictive models of chlorophyll-a, -b and zeaxanthin based on overall dataset, climatic zones (Mediterranean, Continental and Boreal) and depth types (Shallow and Deep) with the best-fitting multivariate model reported. Predictors include total nitrogen (TN), total phosphorus (TP), epilimnetic temperature (TEpi), maximum buoyancy frequency (BuoyFreq), light climate (Zeu/Zmix) and the interaction terms TP * TEpi, TN * TEpi and Zeu/Zmix * BuoyFreq. Linear Model Overall R2 F df Overall dataset Chlorophyll-a (n = 169) 0.23 6.879 161
Log10 Chla = - 6.87 + 1.23 Zeu/Zmix – 2.44 TP + 0.34 TN + 3.18 TEpi – 0.09 BuoyFreq + 0.89 TP * TEpi + 0.27 Zeu/Zmix * BuoyFreq Chlorophyll-b (n = 142) 0.26 11.99 137
Log10 Chlb = -8.19 - 5.05 TN + 2.38 TEpi - 0.24 BuoyFreq + 1.75 TN * TEpi Zeaxanthin (n = 169) 0.21 6.891 162
Log10 Zea = -12.09 + 1.14 Zeu/Zmix - 2.67 TP + 4.01 TEpi – 0.11 BuoyFreq + 0.95 TP * TEpi + 0.21 Zeu/Zmix * BuoyFreq Mediterranean lakes Chlorophyll-a (n = 40) 0.50 5.531 33
Log Chla = - 35.7 + 0.4 Zeu/Zmix -12.4 TP + 12.2 TEpi – 0.11 BuoyFreq + 4.15 TP * TEpi + 0.16 Zeu/Zmix * BuoyFreq Chlorophyll-b (n = 33) 0.73 11.51 26
Log Chlb = - 50.8 – 16.37 TP – 24.87 TN + 15.32 TEpi – 0.32 BuoyFreq + 5.16 TP * TEpi + 7.89 TN * TEpi Zeaxanthin (n = 40) 0.52 7.319 34
Log Zea = - 15.2 + 0.31 TP – 7.15 TN + 4.72 TEpi – 0.14 BuoyFreq + 2.2 TN * TEpi Continental lakes Chlorophyll-a (n = 89) 0.32 7.644 83
Log Chla = - 7.53 + 1.2 Zeu/Zmix + 0.52 TN + 2.81 TEpi - 0.29 BuoyFreq + 0.26 0.21 Zeu/Zmix * BuoyFreq Chlorophyll-b (n = 74) 0.25 5.831 69
Log Chlb = - 11.1 – TN + 3.20 TEpi – 0.34 BuoyFreq + 1.53 TN * TEpi
82 Zeaxanthin (n = 89) 0.13 3.256 84
Log Zea = - 6.22 + 0.16 TP – 0.43 TN + 2.31 TEpi +
0.23 TN * TEpi Boreal lakes Chlorophyll-a (n = 39) 0.67 8.95 31
Log Chla = - 7.62 + 3.86 Zeu/Zmix – 0.39 TP – 16.7 TN + 3.99 TEpi + 0.35 BuoyFreq + 5.7 TN * TEpi + 0.73 Zeu/Zmix * BuoyFreq Chlorophyll-b (n = 34) 0.55 4.533 26
Log Chlb = -9.89 + 2.06 Zeu/Zmix – 0.52 TP – 9.41 TN + 3.05 TEpi + 0.009 BuoyFreq + 3.29 TN * TEpi + 0.47 Zeu/Zmix * BuoyFreq Zeaxanthin (n = 39) 0.39 7.436 35
Log Zea = 2.87 + 4.87 Zeu/Zmix + 0.40 BuoyFreq + 0.89 Zeu/Zmix * BuoyFreq Shallow lakes Chlorophyll-a (n = 74) 0.33 11.62 70 Log Chla = + 4.25 - 0.87 Zeu/Zmix + 0.40 TP + 0.34 TN Chlorophyll-b (n = 64) 0.24 9.79 61 Log Chlb = - 0.27 + 0.80 TN – 0.23 BuoyFreq Zeaxanthin (n = 74) 0.19 8.579 71
Log Zea = + 1.74 - 0.57 Zeu/Zmix + 0.32 TP Deep lakes Chlorophyll-a (n = 95) 0.19 5.256 90
Log Chla = - 0.36 + 2.50 Zeu/Zmix + 1.15 TEpi + 0.19 BuoyFreq + 0.50 Zeu/Zmix * BuoyFreq Chlorophyll-b (n = 78) 0.15 3.267 73
Log Chlb = - 3.35 + 2.02 Zeu/Zmix + 1.38 TEpi + 0.16 BuoyFreq + 0.38 Zeu/Zmix * BuoyFreq Zeaxanthin (n = 95) 0.11 2.854 90
Log Zea = -3.31 + 1.48 Zeu/Zmix + 1.10 TEpi + 0.05 BuoyFreq + 0.27 Zeu/Zmix * BuoyFreq
83 CHAPTER 5 Temperature effects explain continental scale distribution of cyanobacterial toxins
Mantzouki et al. 2018, Toxins
84 Abstract Insight into how environmental change determines the production and distribution of cyanobacterial toxins is necessary for risk assessment. Management guidelines currently focus on hepatotoxins (microcystins). Increasing attention is given to other classes, such as neurotoxins (e.g. anatoxin-a) and cytotoxins (e.g. cylindrospermopsin) due to their potency. Most studies examine the relationship between individual toxin variants and environmental factors, such as nutrients, temperature and light. In summer 2015, we collected samples across Europe to investigate the effect of nutrient and temperature gradients on the variability of toxin production at a continental scale. Direct and indirect effects of temperature were the main drivers of the spatial distribution in the toxins produced by the cyanobacterial community, the toxin concentrations and toxin quota. Generalized linear models showed that a Toxin Diversity Index (TDI) increased with latitude, while it decreased with water stability. Increases in TDI were explained through a significant increase in toxin variants such as MC-YR, anatoxin and cylindrospermopsin, accompanied by a decreasing presence of MC-LR. While global warming continues, the direct and indirect effects of increased lake temperatures will drive changes in the distribution of cyanobacterial toxins in Europe, potentially promoting selection of a few highly toxic species or strains.
Keywords: microcystin; anatoxin; cylindrospermopsin; temperature; direct effects; indirect effects; spatial distribution; European Multi Lake Survey
85 Introduction As a consequence of human population growth, along with associated agricultural, urban and industrial activities, harmful algal blooms worldwide are on the increase (Codd et al. 2017). Eutrophication, one major outcome of anthropogenic activities in the catchments of aquatic ecosystems, is consistently recognized as the main driver of cyanobacterial blooms (Huisman et al. 2005; Conley et al. 2009). In addition, damage to ecosystems and loss of natural resources (e.g. in Lake Taihu, China - Otten et al. 2012 and references within UNEP 2016) are also attributed to on-going climatic change (IPCC 2014). A synergistic interaction between increased nutrients and climate-related changes is predicted (Moss et al. 2011) based on experimental (Lürling et al. 2017) and field studies (Rigosi et al. 2014), potentially further exacerbating the occurrence of cyanobacterial blooms. The long history of cyanobacterial adaptation to a wide range of environmental conditions including extremes (Schopf et al. 2002), supports their successful occurrence in a variety of lake ecosystems. These adaptations come in the form of functional traits such as phosphorus storage, buoyancy regulation, nitrogen fixation and the formation of akinetes (resting spores). Extensive research has linked the prevalence of species with specific functional traits to certain sets of environmental conditions (Carey et al. 2012; Mantzouki et al. 2016). For example, Microcystis aeruginosa can rapidly float up in the illuminated near surface layers under conditions of enhanced water column stability (Ibelings et al. 1991), and through buoyancy regulation gain access to both nutrients at deeper layers and light at the surface (Granf & Oliver, 1982). Toxin production, by the production of hepatotoxins (e.g. microcystins (MCs) and nodularin (NOD)), cytotoxins (cylindrospermopsins (CYN)) and neurotoxins (e.g. anatoxins (ATX)), is another common trait of a large number of cyanobacterial species. Although numerous studies have elucidated the chemical properties, biosynthesis and genetics of the most well-known toxins (Arnaud et al. 2017; Hubert & Fastner 2017; Kokocinski et al. 2017), still little is known why toxins are produced and what determines their presence in field populations. There is evidence that production of these secondary metabolites might provide a competitive advantage for example through providing resistance against grazing (DeMott et al. 1991; Gilbert, 1996; Nogueira et al. 2004) or a physiological benefit e.g. in enhancing nutrient uptake or offering protection against oxidative stress (references within Holland et al. 2013). The abundance of toxins during blooms depends on the presence of toxic strains within the cyanobacterial population (Capelli et al. 2017). Different species have been shown to produce specific toxins or even variants (Bortoli et al. 2014; Cerasino et al. 2017, Meriluoto et al. 2017). However, in toxic strains, the environmental factors controlling expression of the toxin synthetase genes are still a contentious issue (Neilan et al. 2013). Lack of consistency in experimental findings, along with a lack of standardization in surveying and sampling design in field studies, so far has hindered
86 a coherent understanding of how environmental stressors are linked to cyanobacterial toxin production and toxin quota (toxin concentration per cell or unit algal biomass). Various studies have shown contradictory responses of toxin producing taxa to similar environmental parameters. For example, experiments with MC-producing Planktothrix agardhii showed that high nitrogen concentrations (one factor in the study to vary among others like phosphorus, temperature, pH, light) were correlated to high MC production in batch cultures (Sivonen, 1990). But in experiments with Microcystis aeruginosa, N-limited chemostat experiments triggered an increase in MC content, with smaller and faster growing cells being mostly promoted as a response to favourable growing conditions (Long et al. 2001). Similarly, an experiment with ATX producers showed that temperature dependent optimal growth conditions did not necessarily result in higher toxin concentration (Rapala et al. 1993). In this study, while Aphanizomenon cultures proliferated at 30 °C, Dolichospermum (formerly Anabaena) cultures suffered at this high temperature (Rapala et al. 1993). Nevertheless, ATX production was reduced by both tested species (Rapala et al. 1993). A subsequent study on ATX or MC producing strains of Dolichospermum indicated strain-specific responses to temperature and light limited conditions (Rapala et al. 2008). Consequently, there is no easy way to deduce from environmental conditions which toxins will come to dominate in a bloom, nor whether toxin concentrations will be high. Microcystins (MCs), as the largest, best described and most diverse group of cyanobacterial toxins (Neilan et al. 2013), have been the focus of management and mitigation guidelines. For drinking water, the World Health Organisation (WHO) has set a provisional guideline value - maintained at the same level in the EU Drinking Water Directive - of 1 µg/L for MC-LR, a value which is accepted in most countries. In recreational waters, however, there is less consistency, even for MC-LR, and national authorities use a variety of risk assessment schemes and criteria to inform management decisions/practice (Ibelings et al. 2014). MC-LR is the best studied MC variant (Falconer et al. 1992; Fawell, 1993) yet other variants such as MC-YR (Wolf & Frank, 2002), -LW and -LF (Faassen & Lürling, 2013) can also be highly toxic and may contribute significantly to the total MCs in a lake. MC-RR, although reported to be ten times less toxic than MC-LR after intraperitoneal injection in mice (LD50), is one of the most frequently reported toxin in the lakes along with MC-LR and MC-YR (Chorus & Bartram, 1999) and may be more harmful to aquatic biota than MC-LR (Ibelings & Havens, 2008). Good toxicity data are lacking for the vast majority of MC variants (presently more than 250 - Meriluoto et al. 2017). Good toxicological data on other toxins such as CYN and ATX are necessary to include a full spectrum of cyanotoxins in human and ecosystem risk assessment (reviewed in Chorus & Bartram, 1999 & Loftin et al. 2016). Although a fair number of toxin surveys at the national level have been carried out (Chorus, 2005), studies have rarely investigated the spatial distribution of different classes of cyanotoxins at larger geographical scales, encompassing lakes of widely
87 different characteristics and environmental diversity. According to studies (Beniston et al. 2007; Vautard et al. 2014), the climate in Europe is already shifting north (e.g. central European countries will experience longer hot summers, similar to presently in Mediterranean countries) and as such a study on cyanotoxins over a large latitudinal gradient may offer insights into their future distribution. In the European Multi Lake Survey (EMLS), lakes across the continent were sampled once in a snapshot approach, for physical, chemical and biological parameters, during the summer of 2015. Standardized and commonly practiced field protocols along with centralized laboratory analyses for all parameters other than microscopy were undertaken for all samples, avoiding inconsistencies between data in this large dataset. In addition, we addressed our research questions minimizing confounding effects of seasonality, by sampling all lakes during the locally two warmest weeks of summer based on at least 10-year air temperature records. In this study, we investigated how the distribution of toxin concentration and toxin quota were defined by environmental parameters. We hypothesize that in an unusually warm summer - like 2015 was in parts of Europe (Hoy et al. 2016) - temperature, either through direct (surface temperature, epilimnetic temperature) or indirect (water stability expressed as maximum buoyancy frequency) effects, strongly influences the distribution of toxin concentrations and toxin quota. Additionally, we hypothesize that under high temperature stress, the stringent selection of specific well-adapted strains of cyanobacteria reduces toxin diversity, potentially promoting dominance by a few highly toxic variants.
Results 2.1. Toxin distribution on a continental scale In the subset of the 137 EMLS lakes used in our analysis, all 7 toxins analyzed were detected in samples from only 3 lakes, which shows that it is possible but rare to have such a diverse number of toxins present in one lake at one moment in time. The presence of 4, 5 or 6 toxins was found in 34, 26 and 25 lakes, respectively. Finally, 18 lakes had 2 toxins and 13 lakes only 1 toxin. MC variants were found in 93 % of the 137 EMLS lakes (Table 1). MC-YR was the most common of the 5 MCs (in 82 % of the subset). Although the variant MC-dmRR was the rarest variant encountered, it had the highest concentrations compared to any other toxin variant (14.89 µg/L in Polish Lake Syczyńskie). The MC variants MC-LW, MC-LF and MC-LY, and the toxin nodularin (NOD) were also analyzed but they were not included in the analysis as they were too scarce (see materials and methods). The MC-LF was present in two Spanish reservoirs (Abegondo and As Forcadas), while MC-LW and MC-LY were not found in any of the EMLS lakes. Nodularin was found only in two Spanish reservoirs (As Forcadas and Valdecanas) in concentrations close to the limit of quantification (0.007 µg/L).
88 CYN was detected in 39 % of the 137 EMLS lakes (Table 1). One German Lake (Grosser Dabelowsee), three Polish Lakes (Bnińskie, Lusowskie and Probarskie), and two Turkish lakes (Caycoren and Mollakoy) (Figure 1a) had solely CYN, in relatively low concentrations (< 0.05 µg/L, supplemental material). ATX was found in 39 % of the EMLS lakes (Table 1), out of which only one Polish Lake (Dziekanowskie) (Figure 1a) produced the specific toxin exclusively, albeit at very low concentrations 0.002 µg/L. Table 1 Summary of toxin variants (Total microcystin: MC-Tot, microcystin YR: MC-YR, microcystin dmLR: MC-dmLR, microcystin LR: MC-LR, microcystin RR: MC-RR, anatoxin: ATX, cylindrospermopsin: CYN, microcystin dmRR: MC-dmRR) ordered by decreasing number of presence in the investigated 137 EMLS lakes. Prese Limit of Concentration Toxin Variant nt (n Quantification Mean Stdv Range (μg/L) lakes) 1 (μg/L) MC-Tot 127 0-17.18 1.20 2.70 MC-YR 113 0-4.92 0.0050 0.14 0.56 MC-dmLR 108 0-3.16 0.0054 0.15 0.50 MC-LR 93 0-3.97 0.0086 0.20 0.55 MC-RR 67 0-3.31 0.0358 0.20 0.50 ATX 54 0-1.33 0.0004 0.03 0.12 CYN 53 0-2.01 0.0004 0.05 0.20 MC-dmRR 52 0-14.89 0.0489 0.52 1.83 1 limit of quantification (LOQ) of LC-MS/MS method measured for an averaged filtered volume = 250 mL.
89
Figure 1 Percentages of (a) toxin concentrations (μg/L) and (b) toxin quota (μg toxin/ μg chlorophyll-a) of each toxin, of the 137 EMLS lakes used in the analyses. Blue shades correspond to the five microcystin variants (MC-YR; MC-dmLR; MC-LR; MC-RR; MC- dmRR), yellow to cylindrospermospin (CYN) and red to anatoxin (ATX). The radius of the pie charts corresponds to (a) the total toxin concentrations and (b) to the total toxin quota.
90 2.2. Multivariate multiple regression analysis The ordination analysis showed a clear delineation among toxin variants in the EMLS lakes (Figure 2). Lakes with MC-dmLR and MC-dmRR were clustered on the negative- value side of the canonical axis 1, with MC-dmLR occupying the positive-value side of axis 2 and MC-dmRR the negative side of the second axis. Lakes with MC-LR and MC-RR were grouped on the positive-value side of axis 1, and on the positive- and negative-value side of axis 2, respectively. MC-YR demonstrated the highest positive correlation with the canonical axis 2 (r = 0.24). ATX correlated significantly with the negative-value side of axis 2 while CYN correlated negatively with the canonical axis 1. The permutation test confirmed the significance of the canonical analysis (p = 0.001 for axis 1, p = 0.03 for axis 2). The redundancy analysis for toxin concentration and toxin quota data yielded the same results (Table 2). Both the distribution of toxin concentrations and toxin quota were defined by epilimnetic temperature (T_Epi), surface temperature (T_Surf), buoyancy frequency (BuoyFreq) and Secchi depth (Secchi). Therefore, in Figure 2 we present only the biplot for the toxin quota, since the plot for the toxin concentration was almost identical. Marginal tests showed that T_Epi, T_Surf, BuoyFreq and Secchi were all significant in determining the toxin variant ordination (Table 2). T_Epi correlated closely with the negative-value side of the axis 1 (r = -0.62). T_Surf and BuoyFreq had a stronger correlation with the axis 2 (r = -0.55 and r = -0.54). The positive-value side of the second axis correlated strongly with Secchi (r = 0.73). Variance partitioning showed that T_Epi explained 7.3 %, T_Surf 2.5 % and BuoyFreq 1 % of the variance (Figure S 1a), while the Venn diagram on T_Epi, T_Surf and Secchi demonstrated 11, 7 and 1 % of variance explained, respectively (Figure S 1b).
91 Table 2 Redundancy analysis showing results of marginal tests for toxin concentrations followed by toxin quota (both Hellinger transformed) based on F-model and 9999 permutations. Epilimnetic temperature (T_Epi), surface temperature (T_Surf), maximum buoyancy frequency (BuoyFreq) and Secchi depth (Secchi) were the predictors that were selected (stepwise elimination) for the constrained analysis. The Adjusted R2 (AdjR2) estimates the relative quality of the two models. Statistically significant effects are shown in bold. RDA AdjR2 Predictor Variance F p T_Epi 0.05 13.22 0.001 Toxin T_Surf 0.02 4.93 0.002 Concentratio 0.14 ns BuoyFreq 0.01 3.17 0.01 Secchi 0.01 2.87 0.01 T_Epi 0.05 13.22 0.001 T_Surf 0.02 4.93 0.003 Toxin Quota 0.14 BuoyFreq 0.01 3.17 0.02 Secchi 0.01 2.87 0.02
Figure 4 RDA biplot of the toxin quota (toxin μg / chlorophyll-a μg; Hellinger transformed due to many zeros) of the five microcystin variants (MC-YR; MC-dmLR; MC-LR; MC-RR; MC-dmRR), cylindrospermopsin (CYN) and anatoxin (ATX). The vectors represent the environmental variables: epilimnetic temperature (T_Epi), surface temperature (T_Surf) and the log transformed Secchi depth (Secchi) and maximum buoyancy frequency (BuoyFreq). Length and direction of vectors indicate the strength and direction of the relationship.
92 2.3. Toxin diversity index and environmental parameters
The environmental parameters, maximum depth (DMax), latitude (Latitude), epilimnetic temperature (T_Epi), maximum buoyancy frequency (BuoyFreq) and Secchi depth (Secchi), were selected (stepwise selection) as the best explanatory variables in the final model for both the TDI and Richness. The negative binomial generalized linear model showed a significant positive effect of latitude, and a significant negative effect of maximum buoyancy frequency in defining the TDI on a continental scale (Table 3). In the case of Richness, the model showed again a significant positive effect of latitude and a significant negative effect of maximum buoyancy frequency. Additionally, epilimnetic temperature (T_Epi) had also a significant positive effect, while Secchi depth had a significant negative effect in determining Richness. Both of these factors, however, explained less variance than latitude and buoyancy frequency (X2, Table 3). Each toxin quota was tested separately against the TDI to reveal responses in individual toxins to changes in overall toxin diversity (Table 4). The negative binomial generalized linear model showed that the variants MC-YR and MC-dmLR increased significantly (p < 0.05) with increases in TDI. The response of CYN and ATX to increases in the TDI were positive and highly significant (p < 0.01). Positive trends in total microcystin, MC-RR and MC-dmRR were also determined but without any statistical significance (p > 0.05). MC-LR was the only toxin variant that showed a negative trend (red arrow) to increases in toxin diversity, although lacking statistical significance. Similarly, all toxin quota were tested against Toxin Richness. In this case, all toxin quota increased significantly with toxin richness apart from MC-dmLR and MC-LR that although showing a positive trend, were not significant (Table 4).
93 Table 3 Summary of the Generalized Linear Model for the Toxin Diversity Index (TDI) and Toxin Richness of toxin quota. Stepwise elimination selected for final model with predictors maximum depth (DMax), latitude (Latitude), epilimnetic temperature (T_Epi), maximum buoyancy frequency (BuoyFreq) and Secchi depth (Secchi). Statistically significant variables are shown in bold. GLM, family = negative Index Predictor X2 F binomial Latitude 1.21 0.004 BuoyFreq 0.75 0.02 -1.93 + 0.003 DMax + 0.03 TDIquota DMax 0.08 0.8 Latitude** + 0.03 T_Epi T_Epi 0.24 0.2 Secchi 0.30 0.15 Latitude 1.49 0.006 BuoyFreq 2.14 0.001 -0.16 + 0.002 DMax + 0.02 Richnessquota DMax 0.41 0.15 Latitude** + 0.04 T_Epi* T_Epi 1.13 0.02 Secchi 1.40 0.007
94
Figure 3 Map of the Toxin Diversity Index (TDI) of the 137 EMLS lakes, calculated using the Shannon equation. TDI is categorized in four classes with higher colour density (red) representing higher toxin diversity and lower colour density (white) lower toxin diversity. The radius of the markers corresponds to the total toxin concentration in μg/L.
95 Table 4 Statistical results of the negative binomial generalized linear model, showing the response of the toxin quota (MC-YR, MC-dmLR, MC-LR, MC-RR, MC-dmRR, ATX and CYN over chlorophyll-a) to increases in Toxin Diversity Index (TDI) and Toxin Richness. Black upward arrows correspond to increases of the toxin variant to increases in the TDI and Toxin Richness, red downward arrows correspond to decreases of the toxin variant when TDI increases. Statistically significant effects are shown in bold (p < 0.05); highly significant results are marked with 2-3 asterisks (p < 0.01).
Response Response Toxin Quota when TDI Χ2 p when Richness Χ2 p ↑ ↑ MC-YR ↑ 0.10 0.02 ↑ 0.10 0.01 MC-dmLR ↑ 0.10 0.02 ↑ 0.06 0.06 MC-LR ↓ 0.09 0.54 ↑ 0.3 0.2 MC-RR ↑ 0.44 0.06 ↑** 0.90 0.003 0.00 < ATX ↑*** 0.15 ↑*** 0.19 02 0.0001 0.00 CYN ↑** 0.38 ↑*** 0.56 0.0009 7 MC- dmRR ↑ 0.2 0.85 ↑ 0.41 0.02
Discussion Our study shows that MCs were, by far, the most abundant cyanotoxins across the European lakes in our dataset, being detected at greater frequency than either CYN or ATX (Table 1). However, it is important to note that we analyzed only the intracellular toxin content on filter samples, which might have resulted in an underestimation of CYN or ATX concentrations, as they can be largely extracellular (Chorus & Bartram, 1999). We found that among the microcystins, MC-LR was only the third most abundant microcystin variant, after MC-YR and MC-dmLR (Table 1). MC-dmRR was the least common toxin in the EMLS lakes, but showed the highest concentrations (up to 14 μg/L). Furthermore, CYN was detected less frequently but in several cases, it was the only toxin detected which could indicate that CYN producers might have a potential to exclude the producers of other toxin variants (Figure 1). CYN can be present over extended period in aquatic systems, since it can be produced by a succession of different bloom species. For example, in Lake Albano (Italy) a succession in CYN production by Cylindrospermopsis raciborskii to Aphanizomenon ovalisporum lead to the toxin being present in the system from early summer until early autumn (Messineo et al. 2010). ATX also occurred as single toxin, i.e. not in complex mixtures with other cyanotoxins, albeit at lower concentrations (Figure 1). A concrete example
96 is the ongoing substitution of P. rubescens (mainly a MC-dmRR producer) by Tychonema bourrellyi (ATX-producer) in Lake Garda (Italy). Over the last decade, a shift in dominance between these two species caused an increase in ATX at the expense of MC-dmRR (Salmaso et al. 2016). However, and perhaps more typically, there are studies that highlight ATX dominance during short periods of time only, most likely because MC producing taxa take over after a short period of dominance by ATX producers (Pawlik-Skowronska et al. 2013). These results indicate that risk assessment should be broader and address other toxin variants than just the well- known MC-LR variant. A similar conclusion has been reached by other studies, with relevance for human risk assessment but equally for ecosystem functioning (Ibelings & Havens 2008). In the cyanobacterial literature, it is entirely customary to discuss data on presence of toxins in direct relation to the cyanobacterial taxa that produce them. Yet, here we only present data on toxins, no information on the taxonomic composition of the phytoplankton communities of the EMLS lakes is given. Why is this the case? To begin with, this goes back to one of the key-principles underlying the EMLS, one of complete data integration. As explained in the methods nutrients, HPLC pigments or toxins, come from one instrument, operated by one person. Samples for microscopy were taken, but each participating laboratory counted these locally, using different quality microscopes and varying levels of taxonomic expertise. Given a recent discussion (Straile et al. 2013) on problems with the trustworthiness, even at the genus level, of a long-term phytoplankton dataset in which all samples were counted by a small team of experts, supervised by the same person over the years, we could not trustfully use microscopic counts from so many different labs in our study. Moreover, in recent studies there is a strong tendency to focus on key functional traits as the focal point of phytoplankton ecology (e.g. Litchman & Klausmeier, 2008). Cyanobacterial key traits, rather than taxonomic relatedness are also the basis for successful management of cyanobacterial blooms (Loftin et al. 2016). For the purpose of this study, we examined toxins as functional traits and aimed to study how much we can understand about the spatial drivers of toxin abundance by just focusing on the traits themselves. This trait- centred view may be further supported by the fact that all countries base their assessment of the risks of toxic cyanobacteria for the consumption of drinking water or food directly on toxin concentrations and not on taxonomic information (Ibelings et al. 2014). A further EMLS paper is in preparation where we compare traits – pigments from HPLC and size related traits from flowcytometry - with phylogenetic information (16s rRNA) and functional genes (toxin synthetases) to better understand the occurrence of cyanobacteria at the continental scale. Functional trait diversity of cyanobacteria explains the coexistence or succession of the different cyanobacteria species under diverse environmental conditions or lake settings (Carey et al. 2012). However, species-specific toxin production is rarely attributed to environmental factors with certainty. Clusters of genes encoding different cyanotoxin classes can be selectively present in different cyanobacteria
97 species and strains, but mutations can also turn a toxic genotype into non-toxic, under conditions that are not exhaustively studied (Kurmayer et al. 2017). Phylogenetic analysis on the evolutionary age of the MC/NOD synthesis pathways implied that all cyanobacteria are potential MC producers (Rantala et al. 2003). Reported data from Lake Great Prespa (Greece-Macedonia), showed that although certain isolated cyanobacteria species produced more of specific MC variants, they all had the potential to produce all the analyzed MC variants, just in smaller quantities. Toxin concentrations are the result of (i) species abundances, (ii) the abundance of potentially toxigenic genotypes, (iii) the type of toxins that can be produced by those strains, (iv) the cellular quota of the toxins, and finally (v) how all these levels are controlled by environmental settings. In short, the product of toxigenic cyanobacterial biomass x cellular quota determines the toxin concentration in the lake. In terms of environmental drivers, there are those that control cyanobacterial growth and losses (growth – losses = biomass), and most of these are well studied, in particular phytoplankton resources like phosphorus, nitrogen and light (Paerl & Otten, 2013). Moreover, cyanobacterial growth is strongly temperature dependent (Paerl & Huisman, 2008). Population losses are driven by factors like lysis, grazing and parasitism (Visser et al. 2005). Many of the factors that determine biomass have also been found to have an effect on toxin quota (e.g. Long et al. 2001; Wiedner et al. 2003; Jang et al. 2007). Indeed, in line with the overlap in environmental control of biomass and quota from the literature, the distribution of toxin concentrations and quota in EMLS were explained by the same set of environmental predictors. Previous field studies showed elevated MC concentrations in lakes with both low or high cell abundances (Wood et al. 2016). MC concentration per unit biomass can vary considerably from one bloom to another, or even within the same bloom. On the other hand stable toxin quota have also been observed (Hubert & Fastner, 2017). Toxin quota can vary greatly within a toxin producing species, e.g. ranging from 0 - 5 μg/mg of dry cyanobacteria biomass (Sivonen & Jones, 1999). From a management point of view, understanding what drives low toxin quota during high cyanobacterial biomass or high toxin quota during low cyanobacterial biomass versus simply looking at the overall toxin concentrations, would be helpful to better understand variation in toxin concentrations and the risks they pose for use of the water systems (Horst et al. 2014). Even in oligotrophic lakes that typically have low cell densities, the cyanobacterial biomass may accumulate at surface and form scums at leeward shores (Nimtsch et al. 2016), potentially leading to highly localized toxin concentrations, especially when cells possess considerable toxin quota. In contrast, the influence of environmental factors on strain composition is hardly understood. The ordination model showed that temperature effects were mostly responsible for the distribution of the different toxins at a continental scale (Table 2). Interestingly, a significant grouping of lakes with MC-LR and MC-RR on the one hand, and lakes with MC-dmLR and MC-dmRR on the other hand was found (Figure 2). According to these
98 results, lakes with high MC-LR contents would be more likely to have the MC-RR variant as well, while lakes with MC-dmLR are likely to also produce MC-dmRR (Figure 2). Epilimnetic temperature accounted for the delineation of lakes with MC- LR and MC-RR, while lakes with MC-dmLR and MC-dmRR were positively associated with increased buoyancy frequency (Figure 2). This division in lake and toxin groupings indicates that lakes characterized by frequent wind mixing (low buoyancy frequency) and elevated temperature support producers of MC-RR and MC-LR (Mischke, 2003). An example of this type of lake and conditions is Lake Taihu (China), dominated by Microcystis (Otten et al. 2012). Conversely, in deeper and more stable stratified waters, the co-dominance of MC-dmRR and MC-dmLR may be attributed to buoyant species that can accurately regulate their position in the water column providing them a stable position in the metalimnion, like in particular Planktothrix rubescens (e.g. Anneville et al. 2002; Anneville et al. 2004). See below for more in-depth discussion. Our results did not indicate that either total phosphorus or total nitrogen concentrations had a significant impact on the distribution of toxin concentrations or toxin quota (Table 2). As discussed earlier, studies have shown that increased nutrients are linked to increased growth rates and toxin production (references within Neilan et al. 2013 and results of Lürling & Faassen, 2017). However, there is also contradictory evidence that for example, nitrogen availability promoted cell growth, but it did not directly influence toxin production (Sevilla et al. 2010). Hence, there is no consistent evidence supporting a causative relationship between nutrient supported growth and toxin production. In our analyses at the continental scale, nutrients in sharp contrast to temperature effects did not emerge as control factors. We could argue that nutrients would potentially play a role in the occurrence of the individual toxin variants through supporting high cyanobacterial biomasses or toxin quota, but according to the results from our study nutrients would not be the predictor that would select among the different toxin variants, or affect toxin diversity. Toxin diversity and richness showed a significant increase with latitude, which means that northern areas exhibited a higher toxin diversity (Figure 3, Table 4). In a parallel study based on the same lake dataset, Mantzouki et al. (in preparation) showed that during summer 2015 a significantly higher algal - and specifically cyanobacterial - biomass was found in the Boreal climatic zone, compared to lakes in Continental and Mediterranean climate. The higher cyanobacterial biomass was potentially explained by the heat wave that occurred mainly in northern European regions (NOAA online data). The majority of the Boreal EMLS lakes (75 %) were sampled during a two week- period where temperature anomalies exceeded the local long-term average summer temperature by + 5 °C, compared to a much smaller temperature anomaly in the Mediterranean lakes (+ 1.8 °C). Cyanobacteria growth rate steeply increases with water temperature until about 25 °C, and plateaus at about 28 °C, while temperature can be detrimental when it exceeds 33 °C (Paerl, 2014). In 90 % of the Boreal and Continental lakes, with the + 5 °C temperature difference, epilimnetic temperatures
99 approached, but did not exceed 25 °C, giving a higher potential to northern European strains to reach their optimum growth rate. Contrastingly, the Southern European strains were already on the plateau of their growth curves and a 2 °C warming did not add any growth potential. Although a + 5 °C versus a + 2 °C difference may explain why cyanobacteria in the northern Europe, may have caught up with those in the south, it would not explain why cyanobacterial biomass in the north would actually be higher. Clearly, there are differences in temperature responses, both between and within cyanobacterial species (Thomas & Litchman, 2016). The rather extreme + 5 °C temperature anomaly could have altered community composition and/or have favoured cyanobacterial genotypes with an exceptional set of thermal reaction norms (Thomas & Litchman, 2016). The extremely warm summer of 2015 in the north is likely to have selected genotypes which are at the extreme upper - warm - end of the “set of reaction norms” that have evolved locally. This could potentially explained why cyanobacteria in the Boreal lakes developed higher biomass. High buoyancy frequency, as a proxy for water stability, had a significantly negative effect on the TDI and richness of the EMLS lakes (Table 3). In strongly stratified lakes, in particular those with an oligo-mesotrophic state, highly selective conditions arise with a strict spatial separation of light at the surface and nutrients at depth. This leads to the selection of metalimnetic species with a very specific set of functional traits, like well-controlled buoyancy regulation and elevated phycoerythrin content (Anneville et al. 2004). Under such conditions where a single cyanobacterium monopolizes the resources, we may expect that the low cyanobacterial diversity leads to a low toxin diversity (Table 3). On the other hand, in lakes with less stringent environmental conditions, like the more shallow and eutrophic lakes in the EMLS dataset, the scope for co-existence of several less specialised cyanobacterial species is enhanced, hence a more diverse toxin community might be established. In the EMLS dataset, high MC concentrations did not exclude production of the other two toxin classes (neurotoxins, cytotoxins), but rather increased the probability of CYN and ATX occurrence together with MCs, resulting in an increase in the TDI (Table 4). A diverse representation of toxin variants can increase the relative toxic potential of a lake ecosystem (Cerasino & Salmaso, 2012). As the different toxin classes have different modes of action and target different organs, separate toxicity assessments are required (Wolf & Frank, 2002), but ultimately these separate assessments need to be combined to evaluate the overall toxicity risks. The relative toxic potential of the cyanotoxin mixture is calculated as the sum of the relative abundance of each toxin variant multiplied by a defined Toxicity Equivalent Factor based on LD50 values, for each toxin class separately (e.g. neurotoxins vs. hepatotoxins), as proposed in (Wolf & Frank, 2002). As the presence of different toxin classes increases significantly with toxin diversity (Table 4), the differentiated toxic potential would have ramifications for understanding the overall risks of blooms in a lake with an elevated TDI. A higher toxin diversity would potentially lead to higher stability in overall toxicity within a bloom, since if one toxin declines, another may
100 increase, leading to persistence in overall toxicity (Neilan et al. 2013). To make things worse, it may not be sufficient to look at the sum of toxins present (additive effects), since synergistic effects of cyanotoxin mixtures pose a potential risk to humans, animals and aquatic ecosystems (Pires et al. 2011; Freitas et al. 2014; Rzymski et al. 2014). To conclude, we demonstrated that temperature effects were largely responsible for the distribution of the different cyanotoxins on a continental scale. Additionally, we showed that temperature related mechanisms lead to the selective development of well-adapted strains of cyanobacteria that would reduce toxin diversity, potentially promoting dominance by a few highly toxic strains. Further, high buoyancy frequency, as a proxy for water stability, had a significantly negative effect on the TDI and toxin richness. Overall, our study provided the - perhaps surprising - result that at this large-scale temperature rather than classic drivers of toxic blooms like nutrients, determines the distribution of toxins. Materials & Methods 4.1 Sampling survey The European Multi Lake Survey (EMLS) was organized by 26 European countries during summer 2015, with each participating group using their own financial means to conduct their sampling. Since the EMLS was a voluntary effort, individual countries contributed samples from lakes that they routinely sampled, and these were typically lakes with a history of eutrophication. A total of 369 lakes were sampled using standardized protocols for sampling, processing and preserving. Sampling took place during the two warmest weeks of the summer, which was specific for each region. Data collectors identified the correct sampling period using long-term air temperature data spanning at least 10 years. The sampling location was either the historically sampled location, or the centre of the lake if the lake was not routinely sampled. A temperature profile, with a minimum required resolution of 0.5 m sampling intervals, served to define the sampling depth. An integrated water column sample, which will henceforth be referred to as epilimnetic sample, was taken from 0.5 m below the surface until the bottom of the thermocline. This was defined as the point where there was a ≥ 1 °C temperature change per meter lake depth. If the lake was shallow, then the entire water column was sampled until 0.5 m above the lake bottom. All data collectors constructed a simple device using a stoppered hose of the correct length in order to acquire the epilimnetic sample. The hosepipe was lowered with the bottom end open in the water, at a depth of just under the end bottom of the thermocline. When the hosepipe was vertical and the water level was visible at the surface layer of the hosepipe then the stopper was inserted to create vacuum pressure. The bottom end of the hosepipe was pulled to the surface to collect the epilimnetic sample in a bucket. The diameter of the hosepipe was appropriate to sample the required water volume (about 5-10L for hypertrophic and eutrophic, 15-30L for mesotrophic and oligotrophic lakes) for the analyses, in an acceptable number of runs.
101 The first three sampling runs served the purpose of rinsing the hosepipe, the sampling bucket and the plastic rod. The subsequent runs were the water sample taken for analysis. The water sample in the bucket was mixed adequately before being divided into different bottles for further processing prior to analysis. For pigment and toxin analyses, a volume of 50-250 mL for hypertrophic and eutrophic lakes, and 500-1000 mL for mesotrophic to oligotrophic lakes, was filtered through 47mm Glass fibre filters (GF/C or GF/F or similar) using a filtration device. The filters were stored in -20 °C and in the dark until shipping. For analyses of total nutrients, unfiltered water subsamples of 50 mL were stored in -20 °C until shipping. All samples were shipped frozen using dry ice in Styrofoam boxes. All participating countries took part in a one-week training school to discuss and practice all field procedures. All samples were shipped to and stored at the University of Wageningen (Netherlands) until further analysis. Each of the nutrients, pigments and toxins analyses were done in one dedicated laboratory, by one operator on one machine, to minimize analytical errors and maximize integration of the datasets. Specifically, the nutrients, MCs and NOD analyses were done at the University of Wageningen; the pigment analysis at the University of Amsterdam; the CYN and ATX analysis at the German Environment Agency. 4.2 Cyanotoxin analysis In the laboratory, frozen filters were transferred to 8 mL glass tubes and placed for two hours in a freeze-drier (Alpha 1-2 LD, Martin Christ Gefriertrocknungsanlagen GmbH, Osterode am Harz, Germany). The freeze-dried filters then used for the Liquid Chromatography with tandem Mass Spectrometry detection (LC-MS/MS) analysis of microcystins (MCs), nodularin (NOD), cylindrospermopsin (CYN) and anatoxin (ATX) as they are described below. 4.2.1 Microcystins (MCs) and nodularin (NOD) analysis For the extraction of MCs and NOD, 2.5 mL of 75 % hot methanol – 25 % ultrapure water (v/v) was added to the freeze-dried filters, which were then sealed with a screw cap and placed for half an hour at 60 °C. Subsequently, the extract was transferred to a clean 8 mL glass tube. This extraction procedure was performed three times for each filter. The supernatants of the repeated extraction procedure were combined to a final volume of 7.5 mL and then dried in a Speedvac (Thermo Scientific Savant SPD121P, Asheville, NC, USA). After that, the extracts were reconstituted in 900 μL 100 % MeOH. The reconstituted samples were transferred into 2 mL Eppendorf vials with a 0.22 μm cellulose-acetate filter and centrifuged for 5 min at 16.000× g (VWR Galaxy 16DH, Boxmeer, Netherlands). Filtrates were transferred to amber glass vials for the analysis. The LC-MS/MS analysis was performed on an Agilent 1200 LC and an Agilent 6410A QQQ (Waldbronn, Germany). The extracts were separated using a 5μm Agilent Eclipse XDB-C18 (4.6 mm, 150 mm column, Agilent Technologies, Waldbronn,
102 Germany) at 40 °C. The mobile phase consisted of Millipore water (v/v, eluent A) and acetonitrile (v/v, eluent B) both containing 0.1 % formic acid at a flow rate of 0.5 mL/min with the following gradient program: 0–2 min 30 % B, 6–12 min 90 % B, with a linear increase of B between 2 and 6 min and a 5 min post run at 30 % B (as described in [35]). The injection volume was 10 µL. Identification of the eight MC variants (MC- dmRR, MC-RR, MC-YR, MC-dmLR, MC-LR, MC-LY, MC-LW and MC-LF) and nodularin (NOD) was performed in the positive Multiple Reaction Monitoring (MRM) with the following transitions: MC-dmRR 512.8 m/z [M + H]+ to 135.1 quantifier, MC- RR 519.8 m/z [M + H]+ to 135.1 quantifier, MC-YR 523.3 m/z [M + H]+ to 135.1 quantifier, MC-dmLR 491.3 m/z [M + H]+ to 847.6 quantifier, MC-LR 498.3 m/z [M + H]+ to 135.1 quantifier, MC-LY 868.4 m/z [M + H]+ to 163.0 quantifier, MC-LW 891.5 m/z [M + H]+ to 163.0 quantifier, MC-LF 852.5 m/z [M + H]+ to 163.0 quantifier and NOD 825.5 m/z [M + H]+ to 135.1 quantifier (Faassen & Lürling, 2013). Mass spectrometric parameters are given in Faassen & Lürling (2013). Each MC variant was quantified against a calibration curve. The calibration curves were made using certified calibration standards obtained from DHI LAB Products (Hørsholm, Denmark). The limit of detection (LOD) for a 250 mL sample was: 0.0489 µg/l for MC- dmRR, 0.0358 µg/l for MC-RR, 0.0050 µg/l for MC-YR, 0.0054 µg/l for MC-dmLR, 0.0086 µg/l for MC-LR, 0.0817 µg/l for MC-LY, 0.0531 µg/l for MC-LW, 0.0206 µg/l for MC-LF and 0.0048 µg/l for NOD. The limit of quantification (LOQ) for a 250 ml sample was: 0.0489 µg/l for MC-dmRR, 0.0358 µg/l for MC-RR, 0.0050 µg/l for MC-YR, 0.0054 µg/l for MC-dmLR, 0.0086 µg/l for MC-LR, 0.0817 µg/l for MC-LY, 0.0531 µg/l for MC- LW, 0.0206 µg/l for MC-LF and 0.0048 µg/l for NOD. 4.2.2 Cylindrospermopsin (CYN) and anatoxin (ATX) analysis For the extraction of CYN and ATX, 1.5 mL of 0.1% formic acid (FA) was added to the freeze-dried filters, sonicated for 10 min, shaken for 1 hour and then centrifuged. This extraction procedure was repeated two more times and the combined supernatants were dried in a Speedvac (Eppendorf, Germany). Prior to analysis the dried extracts were re-dissolved in 1 mL 0.1% FA and filtered (0.2 µm, PVDF, Whatman, Maidstone, UK). LC-MS/MS analysis was carried out on an Agilent 2900 series HPLC system (Agilent Technologies, Waldbronn, Germany) coupled to a API 5500 QTrap mass spectrometer (AB Sciex, Framingham, MA, USA) equipped with a turbo-ion spray interface. The extracts were separated using a 5 mm Atlantis C18 (2.1 mm, 150 mm column, Waters, Eschborn, Germany) at 30 °C. The mobile phase consisted of water (v/v, eluent A) and methanol (v/v, eluent A) both containing 0.1 % formic acid, and was delivered as a linear gradient from 1% to 25% B within 5 min at a flow rate of 0.25 mL/min. The injection volume was 10 µL. Identification of CYN and ATX was performed in the positive MRM mode with the following transitions: CYN m/z 416.1 [M + H]+ to 194 (quantifier) and 416.1/176, and ATX m/z 166.1 [M + H]+ to 149, 166.1/131, 166.1/91, 166.1/43 (quantifier). Mass spectrometric parameters are given in (Fastner et al. 2018). Certified reference standards were purchased from National Research Council
103 (Ottawa, ON, Canada). The limit of detection (LOD) for both ATX and CYN was 0.0001 µg/l and the limit of quantification (LOQ) was 0.0004 µg/l for a 250 ml sample. 4.3 Nutrient analysis Sample bottles were acid washed overnight in 1M HCl and rinsed with demineralized water before usage. Nutrients were measured using a Skalar SAN+ segmented flow analyser (Skalar Analytical BV, Breda, NL) with UV/persulfate digestion integrated in the system. Total phosphorus and nitrogen were measured in unfiltered subsamples, following Dutch standards protocols (NEN, 1986; NEN, 1990). The limit of detection (LOD) was 0.02 and 0.2 mg/L for total phosphorus and total nitrogen respectively. 4.4. Pigment analysis The analysis of pigments was modified from the method described by (Van der Staay et al. 1992). All filters were freeze dried. Filters (45 mm GF/C and GF/F) were cut in half, placed in separate Eppendorf tubes, and kept on ice until the end of the procedure. We added 600 µL of 90 % acetone to each tube along with a small amount of 0.5 mm beads. To release the pigments from the phytoplankton cells, filters were placed on a bead-beater for one minute. Next, they were placed in an ultrasonic bath for ten minutes to increase the extraction yields. This procedure was repeated twice to ensure a complete extraction of the total pigment content from the filters. To achieve binding of the pigments during the High-Performance Liquid Chromatography (HPLC) analysis, 300 µL of a Tributyl Ammonium Acetate (1.5 %) and Ammonium Acetate (7.7 %) mix were added to each tube. Lastly, samples were centrifuged at 15.000 rpm and 4°C for ten minutes. 35 µL of the supernatant from both Eppendorf tubes of a filter, were transferred into a HPLC glass vials. Pigments were separated on a Thermo Scientific ODS Hypersil column (250 mm × 3 mm, particle size 5 μm) in a Shimadzu HPLC and using a KONTRON SPD-M2OA diode array detector. The different pigments were identified based on their retention time and absorption spectrum and quantified by means of pigment standards. 4.5 Response variables and Environmental parameters Our focal response variables were the toxin variants MC-dmRR, MC-RR, MC-YR, MC- dmLR, MC-LR, CYN and ATX. We also calculated the toxin quota as the ratio of each toxin variant concentration (μg/L) and the chlorophyll-a concentration (μg/L). The latter was used as a proxy for the total phytoplankton biomass. We used the environmental parameters latitude (Latitude), longitude (Longitude), Secchi depth (Secchi), sampling depth (DSampl), maximum depth (DMax), total phosphorus (TP), total nitrogen (TN), surface temperature (T_Surf), epilimnetic temperature (T_Epi), maximum buoyancy frequency (BuoyFreq) and light climate (Zeu/Zmix). Latitude, longitude, secchi depth, sampling depth and temperature profiles were measured directly in the field at all sites. We interpolated all the temperature profiles at a 0.5 m resolution to standardize the data, as most of the profiles were obtained at
104 a higher resolution than the required minimum interval of 0.5 m. From the interpolated profiles, we calculated the epilimnetic temperature as the average temperature from surface until the bottom of the thermocline. The surface temperature value corresponded to the surface temperature. We calculated maximum buoyancy frequency (BuoyFreq) as a metric of stratification strength (Leach et al. 2017). In the rLakeAnalyzer package (Winslow et al. 2016) in R.3.3.3, temperature profiles were used to estimate profiles of buoyancy frequency (N2). N2 is defined as the Brunt–Väisälä equation: N2 = -(g/ρ0) * (δρ(z)/δ(z)), where g is the gravitational acceleration, ρ0 is the density at each depth, and δρ(Z)/ δz is the density gradient. The rlakeAnalyzer uses temperature profiles (in our case of 0.5 m resolution) to determine the density gradients, applying thermodynamic equations specific to freshwater systems (Chen & Millero, 1986). The maximum value of buoyancy frequency generated from the profile was used as an indication of depth where stratification was the strongest.
The ratio (Zeu/Zmix) of euphotic depth (Zeu) to the mixing depth (Zmix) describes the light climate that phytoplankton experience while circulating underwater (Scheffer et al. 1997). We calculated Zeu as Zeu = 2*ZSD (Secchi depth). As Zmix in shallow lakes, we used the sampling depth when N2 was 0, or the top of the metalimnetic depth when stratification was present. In deep lakes, we used the top of the metalimnetic depth. This was calculated as the depth where the steepest density gradient was found (Winslow et al. 2016). We used a total of 137 out of the 369 sampled lakes in order to test our hypotheses. Selection of a smaller subset was justified since in some samples several environmental variables were missing either due to shipping issues or due to deviations from preservation protocols that could affect the integrity of the dataset. Additionally, in order to build the ordination approaches described further down, lakes that had zero concentrations in all toxin variants needed to be excluded to build a meaningful similarity matrix of toxins. Hence, it was necessary to sacrifice a big number of lakes to build a concise dataset where all parameters could be used to select the right model and test the toxin distribution in the EMLS. Samples that were below the limit of detection (LOD), i.e. a toxin signal was detected qualitatively but it was too weak to quantify, were assigned a very small value of half the limit of quantification (LOQ) enabling their inclusion in the analysis. The toxins MC-LF and NOD -which were found only in two lakes- and MC-LY and MC-LW -which were absent from all lakes- were removed from the analysis following the approach of (Legendre & Legendre, 2012) for the most rare species in a dataset. Any statistical results included in this paper correspond to the subset of the 137 EMLS lakes. The Supplementary Material provides a table with information on the total number of lakes where toxin variants were (a) not present (no toxin signal), (b) present and (c) missing, for the 369 EMLS lakes (Table S1). All response variables and environmental parameters, along with the Toxin Diversity Index (TDI) and toxin Richness of the 137- lake subset are provided in the Supplementary Material (Table S2).
105 4.6 Statistical analysis The geographical distribution of the toxin variants (Figure 1a) and their toxin quota (Figure 1b) were mapped with QGIS. We use pie charts to show the percentage of each toxin variant in each sampled lake. To investigate the relationship between the toxin concentration/quota, distribution and the environmental parameters, we used canonical redundancy analysis of principal coordinates (RDA) with permutation test (9999 permutations). Analysis of variation inflation factor (VIF) allowed us to use all sampled environmental variables (as mentioned in section 4.5) to test the relationship between toxin concentrations/quota and environmental variables. Most of the environmental parameters were standardized using log10 transformation (except for surface and epilimnetic temperature). The toxin concentration and toxin quota matrices were standardized by Hellinger transformation (Legendre & Gallagher, 2001). A stepwise elimination of environmental predictors was applied to find the set of parameters that could best explain the ordination of the toxin concentrations/quota. The selected environmental parameters were: surface temperature (T_Surf), epilimnetic temperature (T_Epi), maximum buoyancy frequency (BuoyFreq) and Secchi depth (Secchi). Significance of the ordination was provided performing anova analysis. We used the Shannon-Wiener index to calculate a Toxin Diversity Index (TDI) based on the EMLS toxin quota of 137 lakes. The number of toxins (“Richness”) per lake was also calculated based on the same data. A negative binomial generalized linear model was used to determine the effect of significant environmental parameters on defining the TDI. The same model was used to determine the relation of each toxin variant to TDI. For all the above analyses, we selected the most significant environmental parameters using stepwise selection, based on the Akaike Information Criterion. All statistical analyses were performed in R 3.3.3 (2009) using mainly the vegan (Oksanen et al. 2007) and MASS (Venables & Ripley, 2002) packages.
106 Supplementary Material Table S 1 Number of lakes where toxin variants (a) were not present (no toxin signal), (b) present, (c) missing values in the entire EMLS dataset (n = 369 lakes).
Not Present Present Missing Limit of Toxin Variant 0 μg/L in values Quantification1 (n lakes) (n lakes) (n lakes) (μg/L) MC-YR 165 187 17 0.0050 MC-dmLR 167 185 17 0.0054 MC-LR 200 152 17 0.0086 MC-RR 248 104 17 0.0358 ATX 258 88 23 0.0004 CYN 249 94 26 0.0004 MC- dmRR 271 81 17 0.0489 1 limit of quantification (LOQ) of the LC-MS/MS method measured for an averaged filtered volume = 250 mL. Acknowledgements The authors acknowledge COST Action ES 1105 “CYANOCOST – Cyanobacterial blooms and toxins in water resources: Occurrence impacts and management” and COST Action ES 1201 “NETLAKE – Networking Lake Observatories in Europe” for contributing to this study through networking and knowledge sharing with European experts in the field. Evanthia Mantzouki was supported by a grant from the Swiss State Secretariat for Education, Research and Innovation (SERI) to Bas Ibelings and by supplementary funding from University of Geneva. We thank Clare Ahnlund, Ena Suarez and Irene Gallego for helping out with the Swiss survey. We thank Wendy Beekman and Els J. Faassen for the nutrient and toxin analysis. Author Contributions Evanthia Mantzouki coordinated the EMLS, collected data, curated the dataset, analyzed the data and wrote the manuscript. Bas Ibelings conceived the idea for the EMLS, contributed to discussions throughout the study and to writing of the manuscript. Jutta Fastner and Miquel Lürling performed the toxin analyses and contributed to the writing of the manuscript. Lisette deSenerpont Domis, Sven Teurlincx and Damian Chmura assisted in analysing, interpreting the dataset and in writing the manuscript. The rest of the coauthors were responsible for data collection in lakes in their respective countries, for providing invaluable feedback and for finalizing the manuscript.
107 CHAPTER 6 Opinion: Multi-lake snapshot surveys for lake monitoring, more than a shot in the dark
Mantzouki E., Grossart H.-P., Weyhenmeyer G., Dugan H., deSenerpont Domis L., Brookes J., Skaff N., Doubek J., Rose K., Beklioglu M., Teubner K., Nejstgaard J., Sadro S. and Ibelings BW. Accepted in Front Ecol Evol
108 Abstract Lakes are vital components of the landscape that provide important ecosystem services. They act as sentinels of change, integrating information from atmospheric, terrestrial, and hydrological processes. To support sustainable lake management, lakes must be monitored to provide physical, chemical, and biological information. Monitoring strategies range from long-term time-series on individual lakes to short- term snapshot surveys of up to thousands of lakes from disparate locations. Each monitoring strategy has strengths and weaknesses that may be suited for some purposes but not others. In this opinion paper, we argue that within the spectrum of potential strategies, multi-lake snapshot surveys (MLSS) are one good option, depending on the aims of the survey. The time-scale at which a phenomenon of interest occurs determines the sampling frequency that is needed to capture it. We define snapshot sampling as the acquisition of environmental parameters at intervals that violate the Nyquist sampling criterion, as determined by the Nyquist rate, which is twice the maximum component frequency of the function being sampled. Given the highly dynamic nature of lakes – which can only be captured with automated high- frequency approaches – multi-lake snapshot surveys appear to have obvious limitations. Yet this need not be the case. Since it is simply impossible to monitor thousands of lakes using high-frequency approaches, we provide arguments on how multi-lake snapshot surveys over large geographic distances, covering large environmental gradients, may provide reliable insights into the future effects of environmental change, arguably more so than extrapolation from time series on individual lakes. We identify and discuss a number of strengths of multi-lake snapshot surveys such as their value in: (i) status assessment of freshwater systems and (ii) standardised data across large geographical areas, (iii) cost and time efficiency and (iv) developing prediction models of environmental change by using a space-for- time substitution approach.
Keywords: Multi-Lake Snapshot Surveys, Lake monitoring, Nyquist sampling criterion, space-for-time substitution, phytoplankton ecology
109 Introduction - why do we monitor? Environmental degradation and loss of ecosystem services due to anthropogenic activities are an issue of global concern (Cardinale et al., 2012). Lakes act as effective sentinels of environmental change as they respond to atmospheric, terrestrial, and hydrological processes (Williamson et al., 2008). Understanding lake dynamics can help determine the scale and frequency of occurring changes, establish control measures and maintain ecosystem integrity. Thus, monitoring is necessary, but it is rendered impossible since there are over 117 million lakes globally (Verpoorter et al., 2014). Monitoring strategies that range from long-term time-series on individual lakes to short-term snapshot surveys of up to thousands of lakes from disparate locations serve different purposes and cover different temporal- and spatial-scales of ecological phenomena. For example, phytoplankton dynamics can be driven by long-term environmental change (Monchamp et al., 2016), inter-annual variability (Anneville et al., 2004), seasonal succession (Sommer et al., 2012) and diel changes (Ibelings et al., 1991). To efficiently capture their temporal and spatial variability, the appropriate monitoring strategy needs to be chosen (Table 1). Implementation of long-term monitoring strategies face many challenges. Water quality monitoring programs are usually restricted to priority ecosystems (e.g. socio- economically important or “easier to reach”), creating geographical biases in observations which may not be representative of broader regions or even nearby waterbodies (Ruiz-Jaen and Aide, 2005). Thus, long term monitoring alone is insufficient. To develop a global understanding of environmental response, we need to consider both the sampling frequency and efficiency of monitoring. Combining different monitoring methods may allow studying a good proportion of lakes at temporal- and spatial-scales. Here, we explore the advantages and disadvantages of widely used sampling strategies. We focus on multi-lake snapshot surveys and discuss the limitations of the approach. This strategy allows broad spatial coverage, while remaining affordable. We use mostly phytoplankton examples, because of its rapid response to environmental change (Carpenter et al., 2006). Different monitoring strategies Long-term monitoring from routine (discrete) sampling – typically bi-weekly to monthly - addresses ecosystem change under environmental pressure over time by measuring both coarse and fine-resolution responses (e.g. phytoplankton taxonomy) and environmental drivers (e.g. nutrients) that cannot be sampled with automated or remote sensing approaches. The resulting datasets can elucidate long-term impacts on lakes such as eutrophication (North et al., 2014). Such datasets contributed to developing and validating ecological theories, e.g. the alternative stable state theory (Scheffer and van Nes, 2007), which was successfully implemented in lake restoration
110 programs (Ibelings et al., 2007). Long-term sampling may, however, introduce data inconsistencies over time, due to changes in the sampling protocols, analysis methods and staff employed (Straile et al., 2013). Also, the frequency of routine sampling associated with long-term monitoring does not necessarily assure correct capture of lake processes. Long-term monitoring from automated high-frequency sampling allows characterization of fine-scale temporal dynamics. High-frequency sampling can reveal the build-up and break-down of episodic phytoplankton blooms that cannot be captured with routine sampling (Pomati et al., 2011). Grassroots initiatives like GLEON, support the use of automated high-frequency lake stations worldwide (Weathers et al., 2013). In most cases the characterization of phytoplankton dynamics remains limited to chlorophyll- a measurements from fluorescence sensors. Methods like flowcytometry (Pomati et al., 2011) or image analysis (Sosik and Olson, 2007) are expensive, while data handling requires qualified personnel. Affordable fluorescence probes (e.g. Fluoroprobe- Moldaenke, Germany) that measure pigments of different phytoplankton classes could be an alternative but offer limited taxonomic information to determine community dynamics. Remote sensing provides broad spatial coverage and relatively frequent images. The Landsat satellites have operated since 1972, with a 16-day location-specific revisiting time and spatial resolution of 30-79 m. The newly launched Sentinel satellites have a 5-day revisiting time and spatial resolution of 10-60 m (Toming et al., 2016). The advanced radiometric resolution of Sentinel satellites along with published band ratio algorithms that estimate chlorophyll-a, coloured dissolved organic matter and dissolved organic carbon, make them highly suitable for monitoring lakes (Toming et al., 2016). Remote sensing can, however, be limited by cloud cover (Ibelings et al., 2003), and thus needs to be integrated in a multiplatform monitoring approach (Vos et al., 2003) with airborne based remote sensing and good quality in-situ data for ground truthing. Disparate data. The assembly of multi-lake datasets from disparate sources is flourishing. Disparate data provide a broader representation of environmental change at larger spatial-scales and complementary temporal coverages. International collaborations support such efforts and promote open science to achieve deeper understanding of lake ecosystems globally (Soranno and Schimel, 2014). LAGOS-NE comprises thousands of lakes with diverse geographic conditions and land use histories (Soranno et al., 2017). Disparate data have resulted in important insights into lake functioning (e.g., O’Reilly et al., 2015). Integrating disparate data, however, is a great challenge. Lack of standardization in data protocols and heterogeneity in data formats and units necessitates manual integration (Soranno et al., 2017). Such data inconsistencies should be resolved to successfully attribute environmental change to regional characteristics and not to protocol differences (Moe et al., 2008). Trustworthy databases of disparate data require time and qualified specialists, making it a laborious and costly project (Soranno et al., 2017).
111 Multi-lake Snapshot Surveys (MLSS) sample many lakes across large geographic distances, only once, within a predefined period. We define snapshot sampling as the acquisition of biological, chemical and physical parameters at intervals that violate the Nyquist sampling criterion, as determined by the Nyquist rate, which is twice the maximum component frequency of the function being sampled (Marcé et al., 2016). If for example we study diel re-positioning of algal communities in the water column - which is the outcome of processes that operate on short time-scales - we should sample at hourly intervals (Ibelings et al., 1991). Advantages of the MLSS Status assessment of freshwater systems across large geographical areas. MLSS mostly use standard protocols that minimize sampling effort per lake without sacrificing data quality (Mantzouki and Ibelings, 2018; Pollard et al., 2018). Hence, numerous lakes can be sampled across large geographical areas to frequently assess ecological status (e.g., EU Water Framework Directive, Nordic freshwater inventory - Skjelkvale et al., 2001) and provide ecological understanding. For example, the South American Lake Gradient Analysis (SALGA) investigated the role of temperature on cyanobacterial occurrence in shallow lakes along a latitudinal gradient (Kosten et al., 2012). The National Lake Assessment (NLA) of the US Environmental Protection Agency (US- EPA), sampled over 1000 lakes in 2007 and 2012 (Pollard et al., 2018) to study water quality, food web issues (Doubek & Carey, 2017) and changes over time (Leech et al., 2018). The European Multi-Lake Survey (EMLS) sampled 400 lakes to investigate how temperature and nutrients determine variation in algal and cyanobacterial biomass and toxins (Mantzouki et al., 2018). Standardised data across large geographical areas. MLSS can produce highly comparable datasets, with uniform, synchronic data. Data curators can more easily manipulate the collected data (e.g., outliers’ identification) and perform better quality assurance and control. Thus, data integration can be performed with high fidelity. For complete data integration, data collectors should strictly follow standardized procedures. In the EMLS, representatives from 27 European countries jointly defined the research questions and developed the protocols, during a 3-day training school. The trainees obtained hands-on experience in the agreed protocols and then disseminated the information at the national level. Centralization of key analyses (done by one person on one machine) was also a significant step to assure successful data integration (Mantzouki and Ibelings, 2018). Selection of MLSS lakes is based on sound scientific criteria. The NLA uses a Generalized Random Tessellation Stratified Survey Design (GRTS) which is a spatially-balanced probabilistic design that avoids clumping of sampling locations (Kincaid et al., 2013). MLSS typically engage numerous data collectors that sample many lakes simultaneously. Confounding effects of seasonality can thus be avoided. For example, the EMLS sampled during the locally warmest two-week period to focus
112 on cyanobacterial blooms - a distinct feature of summer phytoplankton (Sommer et al., 2012). Cost and time efficiency is an important advantage of MLSS that can enable global participation and thus investigate landscape-related variation in lakes at large spatial- scale (Sadro et al., 2011). The one-time sampling in a MLSS reduces costs and permits the sampling of numerous lakes. MLSS are particularly suited to grassroots approaches that typically have limited financial means and rely on the motivation and dedication of many scientists from different countries. This low-cost approach allows the participation of researchers and institutes with different levels of funding and equipment, since it does not rely on expensive instrumentation. Because the individual sampling effort in MLSS is not particularly time demanding, numerous environmental parameters can be sampled and analysed at a higher analytical resolution. Thus, MLSS can provide a deeper insight into specific ecological relationships (NLA- and EMLS-related references) which cannot be achieved by high- frequency monitoring strategies. Space-for-time substitution (SfTS). Frequently, MLSS aim to capture environmental differences at geographical gradients to provide insight into impacts of future environmental change. MLSS may use space-for-time substitution (SfTS) (Blois et al., 2013) to study present-day spatial phenomena instead of long-term records that often are unavailable (Pickett, 1989). Sampling numerous lakes is needed for an adequate SfTS. The statistical power generated by sampling many different lakes can overcome the risk of gaining idiosyncratic results from long-term monitoring of only a few lakes. To develop reliable SfTS we need to consider that drivers of temporal change are not necessarily constant across various time-scales. Drivers of large-scale spatial variation rather than of shorter-term temporal variation may be better predictors of long-term climatic change in ecosystems. In grassland communities, geographic rather than temporal variation in annual precipitation and plant community structure better predicted climate-driven changes in precipitation (Adler and Levine, 2007). Temporal drivers of lake change may also differ from spatial drivers, at a short temporal-scale (<20 years) probably because the time-scale (rate and persistence) of change differs in space and time (Weyhenmeyer, 2009). Spatial data may capture the lake’s history over time, i.e. the long-term impact of an environmental predictor but not its short-term impact. For instance, dissolved organic carbon (DOC) and partial pressure of CO2 (pCO2) are related at the spatial-scale (Lapierre and Giorgio, 2012) but fast processes such as flushing-rate can result in a decoupling of the two parameters on a temporal-scale (Nydahl et al., 2017). However, long-term and spatial-scale ice breakup data showed similar patterns of temperature effects on ice-off timing (Weyhenmeyer et al., 2004). Similarly, in 1041 boreal lakes the correlation of chemical variability with increased temperature was consistent across space and time (Weyhenmeyer, 2009). Climate change is emerging as a major driver of both spatial
113 and temporal variation in lake dynamics (Weyhenmeyer, 2009), thus a SfTS may be a suitable solution to predict change. Table 5. Characteristics of the five main lake monitoring strategies (Routine sampling, automated high-frequency sampling, remote sensing, disparate data and snapshot sampling) addressing the scale that they can cover (temporal vs. spatial); the investment in time, money, personnel and equipment; the potential outcome with regards data integration, accuracy, efficiency collaboration and data sharing; and potential caveats.
Automated Routine Remote Disparate Snapshot Rank High Sampling Sensing Data Sampling Frequency
e l Temporal scale
a
c
S Spatial scale
Long-term effort Long-term effort Long-term effort Long-term effort Short-term effort Time
t
n
e Relatively expensive Cheap to expensive Relatively cheap Expensive Relatively cheap
m Money
t
s
e Individual-team Individual-team
v Team effort Team effort Team effort
n effort effort
I Personnel Low - High tech Low - High tech High tech High tech Low tech Equipment Long-term usage Long-term usage Long-term usage Long-term usage Short-term usage
Data Integration
e Accuracy
m
o
c
t
u Efficiency
O International Not necessary Not necessary Necessary Necessary Mostly necessary Collaboration Increasingly open Rarely open access Mostly open access Mostly open access Mostly open access Data Sharing access Human error, Lack of funding, Incomplete data Large spatial Weather conditions Caveats seasonality instrument failure integration coverage required
114 Legend Temporal and Spatial scale – green check marks indicate strategies that we deem to be particularly strong in this respect while orange check marks indicate strategies with a potential to cover temporal or spatial scales. Time – the amount of time required to obtain a comprehensive dataset, based on the research question. Snapshot sampling is attractive based upon this criterion, yielding information at shorter time-scales than most other methods. Money – the funds that the end-users need to invest to build or have access to the dataset. Remote sensing for instance is expensive to get up and running, but for end-users in academia the images are often available at no to low cost. Personnel – the amount of (trained) employees needed to acquire data and maintain meaningful datasets. Is it typically a team effort or could individuals or small groups manage by themselves? Equipment – the type of equipment needed to acquire data in a consistent manner, being technologically advanced or not, and remaining functional for longer or shorter periods. Data Integration – how easily can datasets from different sampling efforts be combined into an integrated dataset? Accuracy – It is hard to award distinctions for this criterion, different methods are appropriate for different types of questions Efficiency – “Bang for the buck”. The amount of scientifically valuable data obtained per unit (monetary) investment. International Collaboration – Is international collaboration essential to create a usable dataset? Data Sharing – Feasibility to publish datasets in an open, publicly accessible format Caveats – potential caveats linked to e.g. research purposes not being clear, funds not being permanently available or confounding effects of seasonality for the detection of long-term trends etc.
Conclusions There are obvious trade-offs between monitoring strategies and no single strategy can provide answers to all research questions, lake management or water governance requirements. An ideal approach might be to organise a yearly MLSS, with both previous and new lakes sampled every year and revisited at a certain time-interval to assess changes in the lake status at the a broad spatial-scale. Additionally, time-series from key lakes could be obtained to develop tailor-made SfTS predictive models. We argue that MLSS, if properly designed and executed, comprise a promising solution for assessing lakes globally, ensuring data integration and engaging researchers, managers, policy makers and citizens (Weyhenmeyer et al., 2017). For a successful MLSS, sampled environmental parameters should be carefully chosen to ensure a reliable SfTS. Numerous lakes, well-spread geographically, should be sampled to cover wide environmental gradients. If the right pre-conditions are met and a standardised sampling plan is established, then MLSS can be an accurate and cost- efficient solution. International, grassroots efforts are increasingly establishing automated high-frequency monitoring stations worldwide. These efforts, along with more MLSS initiatives, could eventually contribute towards a better understanding of both spatial and temporal environmental patterns in lakes.
115 Abbreviations MLSS = Multi-Lake Snapshot Surveys SfTS = Space for time substitution Acknowledgements Evanthia Mantzouki was supported by a grant of the Swiss State Secretariat for Education, Research and Innovation to Bastiaan W. Ibelings for participation in CYANOCOST Action ES 1105 and by supplementary funding from University of Geneva. We acknowledge the Global Lake Ecological Observatory Network (GLEON) for their collaborative spirit and enthusiasm that inspired the idea of the European Multi-Lake survey and thus this opinion paper. Author Contributions EM and BI were responsible for drafting and writing the manuscript, initiating communication with the rest of the co-authors and incorporating changes. The rest of the co-authors provided valuable feedback from personal experience with monitoring strategies and contributed to the correction and synthesis of the manuscript.
116 CHAPTER 7
Conclusions and Perspectives Cyanobacteria are undeniably a global problem that is important to address at – ideally – the global scale. Cyanobacteria benefit from the “effects” of anthropogenic activity, such as nutrient pollution and CO2 – CH4 driven global warming, that facilitate their growth. Cyanobacteria have various functional traits like gas-vesicles, heterocysts or akinetes that allow them to thrive under a wide palette of environmental conditions. Cyanobacteria can also produce different types of toxins that can attack the digestive and the nervous system and inhibit protein synthesis of the higher trophic level organisms. Now, it is true that not all cyanobacterial species have all the above-mentioned functional traits. However, many of the species that have a cosmopolitan distribution, are usually the most notorious ones that can monopolize the available resources and outcompete other algal groups by growing uncontrollably. Understanding the eco-physiological traits of harmful species can reveal the “Achilles heel” of the particular functional group, and thus allow us to design the most efficient methods for a successful control and management of the aquatic environment (Mantzouki et al. 2016).
The main pillar of this doctoral project was the European Multi-Lake Survey (EMLS), a grassroots initiative of 27 countries that gathered a dataset of biological, chemical and physical data from 367 lakes during summer 2015. The participants of the EMLS showed that “where there is a will, there is a way” regardless of financial, cultural or political differences among countries (Chapter 1). This self-funded multi-national project is a clear proof that science, and eventually humanity, can profit from an open and inclusive research approach. Three scientific papers (Chapters 2-4) of this doctoral thesis support the scientific value of the EMLS. The principal aim of the EMLS was to investigate if and how (directly or indirectly) increased nutrients and lake temperatures will promote cyanobacterial growth over other algal competitors at the continental scale. As synergistic interactions between the key environmental stressors - that will potentially promote cyanobacterial occurrence - are to be expected, such patterns were also explored using the EMLS dataset. Enhanced cyanobacterial occurrence increases the chances of cyanotoxin presence in the lakes. The EMLS
117 provided the first large inventory of the geographical distribution of cyanotoxins at the continental scale. This spatially-extended survey contributed in drawing the patterns of cyanobacterial response (both cellular biomass and toxins) to environmental gradients, i.e. in demonstrating if a specific combination of environmental characteristics will trigger a certain type of response across the continent, and in spotlighting the severity of the problem.
Modelling and experimental studies clearly hint at a possible synergistic interaction between increased nutrients and temperature in promoting cyanobacterial blooms and their toxin production, especially in lakes with a eutrophic or hypereutrophic status (Brookes and Carey 2011). A synergistic interaction between two drivers would have a greater effect than the sum of their separate effects and thus it would intensify blooms. Yet, convincing evidence from field observations is seldom given. The EMLS addressed this question by sampling mostly (hyper)eutrophic lakes. Data analysis (Chapter 4) demonstrated a significant synergistic interaction between nutrients and temperature that explained high variation in total algal and cyanobacterial biomass, most commonly in Mediterranean and Boreal lakes. Furthermore, enhanced stratification (water column stability), as an indirect effect of global warming, constitutes a major factor in defining water quality in deep lakes (Schwefel, Gaudard et al. 2016). We found that enhanced water column stability along with light climate - as a main indirect effect of nutrients in eutrophic systems – also interacted synergistically in explaining variation of algal and cyanobacterial biomass. This means that when water column stability is strong, mixing is restricted to the upper water layer. This way, the light climate as the ratio of euphotic depth (Zeu) over mixing depth (Zmix), will improve only for the cells that can maintain their position within the shallower near surface mixed layer. Consequently, those phytoplankton cells will bloom and shade the phytoplankton that will be trapped due to sedimentation, in the deeper, light limited water layers. Under such conditions, cyanobacteria species with functional traits, such as buoyancy regulation, will have a competitive advantage over other photoautotrophic taxa.
Chapter 4 provides important support that cyanobacterial growth will increase in a warmer future. Since water column stability is also expected to increase with higher temperatures (Gerten and Adrian 2002), the predicted interactions will accentuate
118 cyanobacterial blooms, at the continental scale. Our data also strongly suggest that increased lake temperature and nutrients will reinforce mostly cyanobacterial, more so than green algal development. Green algae did not exhibit a significant increase with extreme temperature changes in the Boreal lakes, a pattern that was observed for cyanobacteria. These findings highlight the complexity of mechanisms involved in cyanobacterial bloom development and toxin production, even though water temperature and nutrients are established drivers of blooms.
In addition to determining variation of cyanobacterial biomass in a warmer future, it is also necessary to draw the patterns of how environmental change will determine the distribution of cyanobacterial toxins, in order to back up risk assessment and management (Ibelings et al. 2014). Microcystins (MC), as the most abundantly produced toxins, are the ones mostly targeted in risk assessment and management guidelines. Currently, other toxin types, such as neurotoxins (e.g. anatoxin-a) and cytotoxins (e.g. cylindrospermopsin) are increasingly targeted, due to their high toxic potential. Studies mostly examine the direct relationship between single toxin producers and environmental factors, but disagreement among findings is common (Neilan et al. 2013). The European Multi Lake Survey, as a highly standardized campaign, contributed in creating a consistent dataset of cyanotoxins along the continent to investigate how environmental stressors are linked to cyanobacterial toxin production and toxin quota (toxin concentration per unit algal biomass). In Chapter 5, we used community ecology approaches to examine toxins as a “community” of potentially coexisting toxin types, instead of focusing on each toxin separately. Ordination analysis showed that, direct and indirect effects of temperature were mostly responsible for the distribution of the different toxins at the continental scale. Nutrients did not emerge as a determinant of toxin distribution. Studies have shown that increased nutrients are linked to increased growth rates and toxin production, but they have also shown the opposite (Neilan, Pearson et al. 2013). We would argue that nutrients would potentially play a role in the occurrence of the individual toxin variants, but nutrients would not select among the different toxin variants, or affect toxin diversity at the continental scale.
The toxin diversity (calculated as Toxin Diversity Index using the Shannon equation) increased with latitude demonstrating more diverse toxin mixtures in the Boreal
119 region (Figure 3). This finding is connected to higher cyanobacterial and algal biomass that was found in the North compared to the South in Chapter 4. This would intuitively support the opposite of a notion where “Blooms like it hot”, since the normally cooler Boreal lakes had higher and more diverse toxic blooms than the warmer Mediterranean lakes. Nevertheless, summer 2015 was exceptionally hot in particular up north, where temperature anomalies exceeded the local long-term average summer temperature by + 5 °C, compared to the Mediterranean lakes that exhibited a temperature anomaly of “only” + 1.8 °C (NOAA). Cyanobacteria growth rates steeply increase with water temperature until about 25 °C, plateau at about 28 °C, and collapse above 33 °C (Paerl 2014). Strains in the Boreal lakes had a higher potential to reach their optimal growth rate (Boreal lake temperatures ranged between 20-25 °C), while Mediterranean strains were already at the plateau of their growth. Consequently, a +1.8 °C temperature anomaly probably did not add any growth potential. Responses to temperature change differ both between and within cyanobacterial species. The rather extreme + 5 °C temperature anomaly could have altered community composition and/or have favoured cyanobacterial genotypes that are more responsive to higher temperatures (select from a pre-existing “set of reaction norms” within the cyanobacterial populations – Thomas et al. 2015). We could argue that “Blooms like it hotter than usual”, rather than just hot. This means that lake temperature becomes relevant when heatwaves push water temperature beyond the normal regional temperature. More research is needed to explain this surprising but relevant finding of the EMLS. Thomas et al. (2016), used published data of laboratory experiments from all over the world, to assess the global biogeography of phytoplankton temperature traits. They found that the cyanobacterial optimum growth temperature and the maximum persistence temperature (temperature above which population becomes negative) exhibit far less change along a latitudinal gradient. This could potentially mean that the adaptation capacity of cyanobacteria compared to other algal competitors is not affected by geography and thus the locally normal temperature conditions that could establish a less flexible response to change. If then cyanobacteria do not have different behaviour along different latitudes, they can retain their growth potential every time there is a prolonged temperature anomaly that can favour their development. A confirmation of this hypothesis is needed. We could easily set up experiments where strains from different latitudes would be tested
120 under different temperature regimes. Ideally, a highly standardized multi-laboratory experiment could be performed, where strains would be exchanged between laboratories and tested for the “normally” experienced lake temperatures in each location. The GLEON partners might be interested in undertaking such an effort, and creating a highly standardized experimental dataset of cyanobacterial adaptation to temperature change.
Finally, in Chapter 5 we found that high microcystin concentrations increased the probability of cylindrospermopsin and anatoxin occurrence (high TDI correlates with higher cylindrospermopsin and anatoxin concentrations). This can have great repercussions if heatwaves become more frequent, because toxin mixtures have a higher toxic potential than single toxin variant occurrence (Cerasino and Salmaso 2012). It is also possible that higher toxin diversity would potentially lead to higher stability in overall toxicity within a bloom, since if one toxin declines, another may increase, leading to persistence in overall toxicity. As global warming continues, increased lake temperature and water stability will drive changes in cyanobacterial toxin distribution across Europe, and will potentially promote a few highly toxic species or strains.
The EMLS contributes significantly towards predicting how phytoplankton and cyanobacteria will respond to environmental change. Certain analyses are still pending. For example, DNA analysis could help reveal the community structure of the EMLS samples and explain why certain cyanobacterial toxin types proliferate under certain environment conditions, based on functional traits. Quantitative PCR (qPCR) analysis is planned in order to detect and quantify potential microcystin, nodularin, anatoxin and cylindrospermopsin producers while toxin synthetase genes can be targeted to assess the phytoplankton community biodiversity with High Throughput Sequencing (Kristel Panksepp, Tartu-Estonia). This will be the first laboratory effort that will test and implement the molecular detection methods over a large latitudinal and longitudinal gradient that encompasses lakes with different trophic status and depth. Genetic analysis of phytoplankton community diversity can replace microscopy that usually generates biases over the years and among experts. This approach could help establish pure functional diversity approaches in monitoring water. If such methods can be implemented easily, especially in
121 international multi-lake surveys, they can produce enough evidence to match functional traits - rather than taxonomical information - to environmental perturbation (it doesn’t matter who is there, as long as we know how they will behave under change).
Additionally, the dataset of pigment analysis (HPLC) is currently processed using CHEMTAX (Marju Tamm, Tartu-Estonia) which is a statistical procedure that partitions chlorophyll-a (total microalgal biomass) into the major algal groups and determines the relative and absolute contribution of each group. Flowcytometry samples were also analyzed, that can be used to quantify not only the density of the total phytoplankton community, but also the contributions of cyanobacteria or other algal taxa. This estimation is based on the characterisation of the scattering and pigment fluorescence of individual phytoplankton cells. A wide range (2μm-1mm) of cell size can easily be processed, in a highly automated manner that the manual microscopic analysis could never achieve. The CHEMTAX analysis can be then crosschecked with the flowcytometry and high throughput sequencing data to test for coherence among the methods. Assessing the goodness of each method separately in explaining hypotheses (such as the ones we dealt with in this manuscript) can provide support in using automated methods (HPLC, DNA analysis, flowcytometry) instead of manual methods (microscopy) to simplify and standardize the outcome of large sampling efforts such as the EMLS.
It is essential that the publication of the complete EMLS dataset and methodology will be soon available in the open access journal of Nature Scientific Data (Chapter 3). The rest of the data can be incorporated with the EMLS dataset that is available on the online data repository of the Environmental Data Initiative or on GeoNode (for references see Chapter 3). The GeoNode platform offers great opportunities as the sampled variables can be visualized as geospatial data, reused for creating maps or downloaded and shared through the provided web services. With this platform, ideas can be shared and projects can be initiated more readily. Furthermore, GeoNode may promote future collaboration with various agencies and integration of numerous datasets in the construction of a multidisciplinary platform for global risk assessment of aquatic ecosystems. This way the EMLS becomes an example towards achieving Sustainable Development Goals, one of the main targets of the United Nations the last
122 years. The aim is to achieve globally agreed goals by merging international policies with solutions at national and local levels. As the EMLS consists of countries with their own history, legislation and culture for the management of national water resources, it creates the opportunity to establish novel, applied socio-ecological system research, which will greatly aid management and policymaking. We hope that the first EMLS products might inspire similar initiatives to study across large geographic areas and gather more evidence on how lakes respond in a changing environment. We highly encourage scientists to make the best use of the publicly available EMLS dataset and explore other interesting hypotheses that we have not yet addressed. Good water quality is vital for everyone, and if we are to better understand the threats to aquatic ecosystems in a changing world, we must make it our responsibility to use all the data that are out there and generate all possible ideas using our scientific expertise.
Figure 1. A summary of the most important contributions of the European Multi Lake Survey and this dissertation.
123 References
A
Adler PB and Levine JM (2007). Contrasting relationships between precipitation and species richness in space and time. Oikos 116:221-232.
Adrian R et al. (2009). Lakes as sentinels of climate change. Limnology and Oceanography 54:2283-2297.
Agawin N, Rabouille S, Veldhuis M, Servatius L, Hol S, van Overzee HM, Huisman J (2007). Competition and facilitation between unicellular nitrogen-fixing cyanobacteria and non-nitrogen-fixing phytoplankton species. Limnology and Oceanography 52:2233-2248.
Agnihotri VK (2013). Anabaena flos-aquae Critical Reviews in Environmental Science and Technology.
Anagnostidis K, Komárek J (1988). Modern approach to the classification system of cyanophytes. 3-Oscillatoriales. Arch Hydrobiol Suppl 80, Algological Studies 50-53: 327–472.
Anneville O, Gammeter S, Straile D (2005). Phosphorus decrease and climate variability: mediators of synchrony in phytoplankton changes among European peri- alpine lakes. Freshwater Biology 50:1731-1746.
Anneville O, Molinero JC, Souissi S, Gerdeaux D (2010). Seasonal and interannual variability of cladoceran communities in two peri-alpine lakes: uncoupled response to the 2003 heat wave. Journal of Plankton Research 32:913-925.
Anneville O, Souissi S, Gammeter S, Straile D (2004). Seasonal and inter-annual scales of variability in phytoplankton assemblages: comparison of phytoplankton dynamics in three peri-alpine lakes over a period of 28 years. Freshwater Biology 49:98-115.
124 Anneville O, Souissi S, Ibanez F, Ginot V, Druart JC, Angeli N (2002) Temporal mapping of phytoplankton assemblages in Lake Geneva: Annual and interannual changes in their patterns of succession. Limnology and Oceanography 47:1355-1366.
Anneville O, Domaizon I, Kerimoglu O, Rimet F, Jacquet S (2015). Blue-green algae in a “Greenhouse century”? New insights from field data on climate change impacts on cyanobacteria abundance. Ecosystems 18(3):441-458.
Arnaud C, Bernard C, Spoof L, Bruno, M (2017). Microcystins and Nodularins In Handbook of Cyanobacterial Monitoring and Cyanotoxin Analysis Eds: Meriluoto J, Codd GA. John Wiley & Sons, Ltd.
B
Barnosky AD et al. (2012). Approaching a state shift in Earth's biosphere. Nature 486:52-58.
Bauersachs T, Stal LJ, Grego M, Schwark L (2014). Temperature induced changes in the heterocyst glycolipid composition of N-2 fixing heterocystous cyanobacteria. Organic Geochemistry 69:98-105.
Beaulieu M, Pick F, Gregory-Eaves I (2013). Nutrients and water temperature are significant predictors of cyanobacterial biomass in a 1147 lakes data set. Limnology and Oceanography 58(5): 1736-1746.
Beier U, Degerman E, Melcher A, Rogers C, Wirlöf H (2007). Processes of collating a European fisheries database to meet the objectives of the European Union Water Framework Directive. Fisheries Management and Ecology 14:407-416.
Beklioglu M, Tan CO (2008). Restoration of a shallow Mediterranean lake by biomanipulation complicated by drought. Fundamental and Applied Limnology / Archiv für Hydrobiologie 171(2): 105-118.
125 Beniston M et al. (2007). Future extreme events in European climate: an exploration of regional climate model projections. Climatic Change 81(S1):71-95.
Berger C (1975). Occurrence of Oscillatoria agardhii Gomin some shallow eutrophic lakes. Internationale Vereinigung für theoretische und angewandte Limnologie: Verhandlungen, 19(4):2689-2697.
Blois J, Williams JW, Fitzpatrick MC, Jackson ST and Ferrier S (2013). Space can substitute for time in predicting climate-change effects on biodiversity. PNAS 110(23): 9374-9379.
Bootsma, HA (2018). Oceans, lakes, and inland seas: a virtual issue on the large lakes of the world. L&O Bulletin doi:10.1002/lob.10230.
Bormans M, Condie S (1998). Modelling the distribution of Anabaena and Melosira in a stratified river weir pool. Hydrobiologia 364:3-13.
Bormans M, Sherman BS, Webster IT (1999). Is buoyancy regulation in cyanobacteria an adaptation to exploit separation of light and nutrients? Marine and Freshwater Research 50:897-906.
Bormans M, Ford PW, Fabbro L, Hancock G (2004). Onset and persistence of cyanobacterial blooms in a large impounded tropical river. Australia Marine and Freshwater Research 55:1-15.
Bormans M, Ford PW, Fabbro L (2005). Spatial and Temporal variability in cyanobacterial populations controlled by physical processes. Journal of Plankton Research 27(1):61-70.
Bortoli S, Oliveira-Silva D, Krüger T, Dörr FA, Copelicolo P, Volmer DA, Pinto E (2014). Growth and microcystin production of a Brazilian Microcystis aeruginosa strain (LTPNA 02) under different nutrient conditions. Revista Brasileira de Farmacognosia 24(4):389-398.
126 Brentrup JA et al. (2016). The potential of high-frequency profiling to assess vertical and seasonal patterns of phytoplankton dynamics in lakes: An extension of the Plankton Ecology Group (PEG) model. Inland Waters 6(4):565-580.
Briand E et al. (2008). Temporal variations in the dynamics of potentially microcystin-producing strains in a bloom-forming Planktothrix agardhii (Cyanobacterium) population. Applied and Environmental Microbiology 74(12): 3839-3848.
Briand JF, Leboulanger C, Humbert JF, Bernard C, Dufour P (2004). Cylindrospermopsis raciborskii (Cyanobacteria) invasion at mid-latitudes: Selection, wide physiological tolerance, or global warming? Journal of Phycology 40:231-238.
Brookes JD, Carey CC (2011). Resilience to Blooms. Science 334(6052): 46
Butterwick C, Heaney SI, Talling JF (2005). Diversity in the influence of temperature on the growth rates of freshwater algae, and its ecological relevance. Freshwater Biology 50:291-300.
C
Camacho A (2006). On the occurrence and ecological features of deep chlorophyll maxima (DCM) in Spanish stratified lakes. Limnetica 25(1-2): 453-478.
Capelli C, Ballot A, Cerasino L, Papini A, Salmaso N (2017). Biogeography of bloom- forming microcystin producing and non-toxigenic populations of Dolichospermum lemmermannii (Cyanobacteria). Harmful Algae 67:1-12.
Cardinale BJ et al. (2011). The functional role of producer diversity in ecosystems. American journal of botany 98:572-592.
Cardinale BJ et al. (2012). Biodiversity loss and its impact on humanity. Nature 486:59-67.
127 Carey CC, Ibelings BW, Hoffmann EP, Hamilton DP, Brookes JD (2012). Eco- physiological adaptations that favour freshwater cyanobacteria in a changing climate. Water Research 46: 1394-1407.
Carey CC et al. (2012). Occurrence and toxicity of the cyanobacterium Gloeotrichia echinulata in low-nutrient lakes in the northeastern United States. Aquatic Ecology 46:395-409.
Carmichael W (1997). The cyanotoxins. Advances in botanical research 27:211-256.
Carmichael WW (2001). Health effects of toxin-producing cyanobacteria: "The CyanoHABs". Human and Ecological Risk Assessment 7:1393-1407.
Carpenter SR, Bennett EM, Peterson GD (2006). Scenarios for ecosystem services: an overview. Ecology and Society 11(1).
Carpenter SR et al. (2007). Understanding regional change: A comparison of two lake districts. Bioscience 57:323-335.
Castro D, Vera D, Lagos N, Garcı́a C, Vásquez M (2004). The effect of temperature on growth and production of paralytic shellfish poisoning toxins by the cyanobacterium Cylindrospermopsis raciborskii. Toxicon 44:483-489.
Cerasino L, Salmaso N (2012). Diversity and distribution of cyanobacterial toxins in the Italian subalpine lacustrine district. Oceanological and Hydrobiological Studies 41: 54-63.
Cerasino L, Capelli C, Salmaso N (2017). A comparative study of the metabolic profiles of common nuisance cyanobacteria in southern perialpine lakes. Advances in Oceanography and Limnology 8(1).
Chen-Thung A, Millero FJ (1986). Thermodynamic properties for natural waters covering only the limnological range. Limnology and Oceanography 31:657-662.
128 Cheung MY, Liang S, Lee J (2013). Toxin-producing cyanobacteria in freshwater: A review of the problems, impact on drinking water safety, and efforts for protecting public health. Journal of Microbiology 51:1-10.
Chislock MF, Sharp KL, Wilson AE (2014). Cylindrospermopsis raciborskii dominates under very low and high nitrogen-to-phosphorus ratios. Water Research 49:207-214.
Chonudomkul D, Yongmanitchai W, Theeragool G, Kawachi M, Kasai F, Kaya K, Watanabe MM (2004). Morphology, genetic diversity, temperature tolerance and toxicity of Cylindrospermopsis raciborskii (Nostocales, Cyanobacteria) strains from Thailand and Japan. FEMS Microbiology Ecology 48:345-355.
Chorus I, Bartram J (1999). Toxic cyanobacteria in water: A guide to their public health consequences, monitoring and management. Spon Press.
Chorus I (2005). Current approaches to cyanotoxin risk assessment, risk management and regulations in different countries. In Chorus (eds): Federal Environmental Agency (Umweltbundesamt).
Christensen MR, Graham MD, Vinebrooke RD, Findlay DL, Paterson MJ, Turner MA (2006). Multiple anthropogenic stressors cause ecological surprises in boreal lakes. Global Change Biology 12:2316-2322.
Codd GA, Morrison LF, Metcalf JS (2005). Cyanobacterial toxins: risk management for health protection. Toxicology and Applied Pharmacology 203:264-272.
Codd GA, Meriluoto J, Metcalf JS (2017). Introduction: cyanobacteria, cyanotoxins, their human impact, and risk management In Handbook of cyanobacterial monitoring and cyanotoxin analysis. John Wiley & Sons, Ltd.
Conley DJ, Paerl HW, Howarth RW, Boesch DF, Seitzinger SP, Havens KE, Likens GE (2009). Controlling Eutrophication: Nitrogen and Phosphorus. Science 323(5917): 1014.
129 COST European Cooperation in Science & Technology : Cost Actions – http://www.cost.eu/COST_Actions (2017).
Cottingham KL, Ewing HA, Greer ML, Carey CC, Weathers KC (2015). Cyanobacteria as biological drivers of lake nitrogen and phosphorus cycling. Ecosphere 6(1).
CYANOCOST COST Action ES1105 : Cyanobacterial blooms and toxins in water resources: Occurrence, impacts and management – http://www.cyanocost.com/ (2014).
D
Dai A, Trenberth KE, Qian TT (2004). A global dataset of Palmer Drought Severity Index for 1870-2002: Relationship with soil moisture and effects of surface warming. Journal of Hydrometeorology 5:1117-1130.
Davis TW, Berry DL, Boyer GL, Gobler CJ (2009). The effects of temperature and nutrients on the growth and dynamics of toxic and non-toxic strains of Microcystis during cyanobacteria blooms. Harmful algae 8:715-725.
Davy R, Esau I (2016). Differences in the efficacy of climate forcings explained by variations in atmospheric boundary layer depth. Nature communications 7:11690. de Marsac NT (1991). Chromatic adaptation by cyanobacteria. In: Bogorad L, Vasil IK (eds) The photosynthetic apparatus: molecular biology and operation. Academic Press 417-444.
DeMott WR, Zhang QX, Carmichael WW (1991). Effects of toxic cyanobacteria and purified toxins on the survival and feeding of a copepod and three species of Daphnia. Limnology and Oceanography 36(7):1346-1357.
De Tezanos Pinto P, Litchman E (2010a). Eco-physiological responses of nitrogen- fixing cyanobacteria to light. Hydrobiologia. 639:63-68.
130 De Tezanos Pinto P, Litchman E (2010b). Interactive effects of N:P ratios and light on nitrogen-fixer abundance. Oikos. 119:567-575.
De Senerpont Domis, LN, Elser, JJ, Gsell, AS, Huszar, VLM, Ibelings, BW, Jeppesen, E, Lürling, M (2013). Plankton dynamics under different climatic conditions in space and time. Freshwater Biology 58(3):463-482.
Dodds WK et al. (2009). Eutrophication of US freshwaters: analysis of potential economic damages. Environmental Science & Technology 43:12-19.
Dokulil MT, Teubner K (2000). Cyanobacterial dominance in lakes. Hydrobiologia 438(1-3):1-12.
Dokulil MT, Teubner K (2012). Deep living Planktothrix rubescens modulated by environmental constraints and climate forcing. Hydrobiologia 698:29-46.
Dolman AM, Rucker J, Pick FR, Fastner J, Rohrlack T, Mischke U, Wiedner C(2012). Cyanobacteria and cyanotoxins: the influence of nitrogen versus phosphorus. PLoS One 7(6):e38757.
Doubek JP, Carey CC and Cardinale BJ (2015). Anthropogenic land use is associated with N-fixing cyanobacterial dominance in lakes across the continental United States. Aquatic sciences 77:681-694.
Downing JA, Watson SB, McCauley E(2001). Predicting cyanobacteria dominance in lakes. Canadian Journal of Fisheries and Aquatic Sciences 58(10):1905-1908.
Dudgeon D et al. (2006). Freshwater biodiversity: importance, threats, status and conservation challenges. Biological Reviews 81:163-182.
Durden JM, Murphy K, Jaeckel A, Van Dover CL, Christiansen S, Gjerde K, Ortega A and Jones DOB (2017). A procedural framework for robust environmental management of deep-sea mining projects using a conceptual model. Marine Policy 84:193-201.
131 Dziallas C, Grossart H-P(2011). Increasing oxygen radicals and water temperature select for toxic Microcystis sp. PLoS One, 6(9):e25569.
E
Elliott JA, Thackeray SJ, Huntingford C, Jones RG (2005). Combining a regional climate model with a phytoplankton community model to predict future changes in phytoplankton in lakes. Freshwater Biology 50:1404-1411.
Elliott, JA(2012). Is the future blue-green? A review of the current model predictions of how climate change could affect pelagic freshwater cyanobacteria. Water Research 46(5):1364-1371.
Ernst B, Hoeger SJ, O’Brien E, Dietrich DR (2009). Abundance and toxicity of Planktothrix rubescens in the pre-alpine Lake Ammersee Germany. Harmful Algae 8:329-342.
F
Faassen EJ, Lürling, M (2013). Occurrence of the microcystins MC-LW and MC-LF in Dutch surface waters and their contribution to total microcystin toxicity. Marine Drugs 11(7):2643-54.
Falconer IR, Choice A, Hosja W (1992). Toxicity of edible mussels (Mytilus edulis) growing naturally in an estuary during a water bloom of the blue-green alga Nodularia spumigena. Environmental Toxicology and Water Quality 7(2):119-123.
Fastner J et al. (2003) .Cylindrospermopsin occurrence in two German lakes and preliminary assessment of toxicity and toxin production of Cylindrospermopsis raciborskii (Cyanobacteria) isolates. Toxicon 42(3): 313-321.
Fastner J et al. (2018) Fatal neurotoxicosis in dogs associated with Tychoplanktic, Anatoxin-a producing Tychonema spin mesotrophic lake Tegel, Berlin. Toxins 10.
132 Fawell JK, James CP, James HA (1993). Toxins from blue-green algae: toxicological assessment of microcystin-LR and a method for its determination in water; foundation for water research: Medmenham. Bucks, UK, Marlow. 1-46.
Fietz S, Nicklish A (2002). Acclimation of the diatom Stephanodiscus neoastraea and the cyanobacterium Planktothrix agardhii to simulated natural light fluctuations. Photosynthesis Research 72(1): 95-106.
Folt CL, Chen CY, Moore MV, Burnaford J (1999). Synergism and antagonism among multiple stressors. Limnology and Oceanography 44(3):864-877.
Foy RH, Gibson CE, Smith RV (1976). The influence of day length, light intensity and temperature on the growth rates of planktonic blue-green algae. British Phycological Journal 11:151-163.
Fraisse et al. (2013). Morphofunctional traits reflect differences in phytoplankton community between rivers of contrasting flow regime. Aquatic ecology 47:315–327.
Freitas EC, Pinheiro C, Rocha O, Loureiro S (2014). Can mixtures of cyanotoxins represent a risk to the zooplankton? The case study of Daphnia magna Straus exposed to hepatotoxic and neurotoxic cyanobacterial extracts. Harmful Algae 31:143-152.
G
Gallina N, Anneville O, Beniston M (2011). Impacts of extreme air temperatures on cyanobacteria in five deep peri-Alpine lakes. Limnology and Oceanography 70:186- 196.
Gallon JR, Chit KN, Brown EG (1990). Biosynthesis of the tropane-related cyanobacterial toxin anatoxin-a: Role of ornithine decarboxylase. Phytochemistry 29:1107-1111.
133 Ganf GG, Oliver RL (1982). Vertical separation of light and available nutrients as a factor causing replacement of green-algae by blue-green-algae in the plankton of a stratified lake. Journal of Ecology 70:829-844.
Ger KA, Urrutia-Cordero P, Frost PC, Hansson LA, Sarnelle O, Wilson AE, Lürling M (2016). The interaction between cyanobacteria and zooplankton in a more eutrophic world. Harmful Algae 54:128-144.
Gerten D, Adrian R (2002). Effects of climate warming, North Atlantic Oscillation, and El Nino-Southern Oscillation on thermal conditions and plankton dynamics in northern hemispheric lakes. The Scientific World Journal 2: 586-606.
Gkelis S, Zaoutsos N (2014). Cyanotoxin occurrence and potentially toxin producing cyanobacteria in freshwaters of Greece: A multi-disciplinary approach. Toxicon 78:1- 9.
Gilbert, JJ (1996) Effect of food availability on the response of planktonic rotifers to a toxic strain of the cyanobacterium Anabaena flos-aquae. Limnology and Oceanography 41(7):1565-1572.
Glazer AN (1984). Phycobilisome a macromolecular complex optimized for light energy transfer. Biochimica et Biophysica Acta - Reviews on Bioenergetics 768:29-51.
Groemping U(2006). Relative Importance for Linear Regression in R: The Package relaimpo. Journal of Statistical Software 1(1).
Gsell AS, de Senerpont Domis LN, Przytulska-Bartosiewicz A, Mooij WM, van Donk E, Ibelings BW (2012). Genotype-by-temperature interactions may help to maintain clonal diversity in Asterionella formosa (Bacillariophyceae). Journal of Phycology 48(5):1197-1208.
Gu B, Alexander V (1993). Estimation of N2 fixation based on differences in the natural abundance of 15N among freshwater N2-fixing and non-N2-fixing algae. Oecologia 96:43-48.
134 Gulati RD, Pires LMD, Van Donk E (2008). Lake restoration studies: failures, bottlenecks and prospects of new ecotechnological measures. Limnologica-Ecology and Management of Inland Waters 38(3):233-247.
H
Havens KE (2008). Cyanobacteria blooms: effects on aquatic ecosystems In: (ed) Hudnell HK Cyanobacterial harmful algal blooms: state of the science and research needs. Advances in Experimental Medicine and Biology 619:733-747.
Helsel, DR (2006). Fabricating data: How substituting values for nondetects can ruin results, and what can be done about it. Chemosphere 65:2434-2439.
Hillebrand H, Matthiessen B (2009). Biodiversity in a complex world: consolidation and progress in functional biodiversity research. Ecology letters 12:1405-1419.
Holland A, Kinnear S (2013). Interpreting the possible ecological role(s) of cyanotoxins: compounds for competitive advantage and/or physiological aide? Marine Drugs 11(7):2239-58.
Horst GP, Sarnelle O, White JD, Hamilton SK, Kaul RB, Bressie JD (2014). Nitrogen availability increases the toxin quota of a harmful cyanobacterium, Microcystis aeruginosa. Water Research 54:188-98.
Hosper SH (1998). Stable states, buffers and switches: an ecosystem approach to the restoration and management of shallow lakes in the Netherlands. Water Science and Technology 37:151-164.
Hoy A, Hänsel S, Skalak P, Ustrnul Z, Bochnicek O (2016). The extreme European summer of 2015 in a secular perspective. International Journal of Climatology 37(2):1-34.
135 Huber V, Wagner C, Gerten D, Adrian R (2012). To bloom or not to bloom: contrasting responses of cyanobacteria to recent heat waves explained by critical thresholds of abiotic drivers. Oecologia 169:245-256.
Hubert JF, Fastner J (2017). Ecology of Cyanobacteria. In Meriluoto J, Spoof L, Codd GA (eds) Handbook of cyanobacterial monitoring and cyanotoxin analysis. John Wiley & Sons, Ltd.
Hudnell HK, Dortch Q (2008). Chapter 2: A synopsis of research needs identified at the interagency, international symposium on cyanobacterial harmful algal blooms (ISOC-HAB) In: Hudnell HK (ed) Cyanobacterial harmful algal blooms: state of the science and research needs. Advances in Experimental Medicine and Biology 619:17- 43.
Huisman J, Sharples J, Stroom JM, Visser PM, Kardinaal WEA, Verspagen JMH, Sommeijer B (2004). Changes in turbulent mixing shift competition for light between phytoplankton species. Ecology 85(11): 2960-2970.
Huisman J,Matthijs HCP,Visser PM,Ibelings BW,Mur LR, Walsby, AE (2005). The ecophysiology of the harmful cyanobacterium Microcystis. Harmful Cyanobacteria 109-142.
Huisman J, Matthijs HC, Visser PM (2005). Harmful cyanobacteria Springer (3).
Huisman J,Codd GA,Paerl HW, Ibelings BW, Verspagen JMH, Visser PM(2018). Cyanobacterial Blooms. Nature 16:471-483.
Humphries SE, Lyne VD (1988). Cyanophyte blooms: The role of cell buoyancy. Limnology and Oceanography 33:79-91.
136 I
Ibelings BW, Mur LR, Kinsman R, Walsby A (1991). Microcystis changes its buoyancy in response to the average irradiance in the surface mixed layer. Archiv für Hydrobiologie 120:385-401.
Ibelings BW, Kroon BMA, Mur LR (1994). Acclimation of Photosystem-II in a cyanobacterium and a eukaryotic green-alga to high and fluctuating photosynthetic photon flux densities, simulating light regimes induced by mixing in lakes. New Phytologist 128:407-424.
Ibelings B, Admiraal W, Bijkerk R, Ietswaart T, Prins H (1998). Monitoring of algae in Dutch rivers: does it meet its goals? Journal of applied phycology 10:171-181.
Ibelings BW, Maberly SC (1998). Photoinhibition and the availability of inorganic carbon restrict photosynthesis by surface blooms of cyanobacteria. Limnology and Oceanography 43:408-419.
Ibelings BW, Vonk M, Los HFJ, van der Molen DT, Mooij WM (2003). Fuzzy modelling of cyanobacterial surface water blooms: Validation with NOAA-AVHRR satellite images. Ecological Applications 13:1456-1472.
Ibelings BW, Portielje R, Lammens EH, Noordhuis R, van den Berg MS, Joosse W, Meijer ML (2007). Resilience of alternative stable states during the recovery of shallow lakes from eutrophication: Lake Veluwe as a case study. Ecosystems 10:4-16.
Ibelings BW, Chorus I (2007). Accumulation of cyanobacterial toxins in freshwater "seafood" and its consequences for public health: A review. Environmental Pollution 150:177-192.
Ibelings B, Havens, K (2008). Cyanobacterial toxins: a qualitative meta-analysis of concentrations, dosage and effects in freshwater, estuarine and marine biota. Advances in Experimental Medicine and Biology 619:675-732.
137 Ibelings BW, Backer LC, Kardinaal WEA, Chorus I (2014). Current approaches to cyanotoxin risk assessment and risk management around the globe. Harmful Algae 40: p63-74.
IPCC Climate Change (2014). Impacts, adaptation and vulnerability. Contribution of working group ii to the fourth assessment report of the Intergovernmental Panel on Climate Change. In (Eds) Field CB, Barros VR, Dokken DJ, Mach KJ, Mastrandrea MD, Bilir TE, White LL. Cambridge, United Kingdom and New York, NY, USA.
Istvanovics V, Shafik H, Presing M (2000). Growth and phosphate uptake kinetics of the cyanobacterium Cylindrospermopsis raciborskii (Cyanophyceae) in through flow cultures. Freshwater Biology 43:257-275.
J
Jacquet S et al. (2005). The proliferation of the toxic cyanobacterium Planktothrix rubescens following restoration of the largest natural French lake (Lac du Bourget). Harmful Algae 4:651-672.
Jang MH, Jung JM, Takamura N (2007). Changes in microcystin production in cyanobacteria exposed to zooplankton at different population densities and infochemical concentrations. Limnology and Oceanography 52(4):1454-1466.
Jeppesen E, Kristensen P, Jensen JP, Søndergaard M, Mortensen E, Lauridsen T (1991). Recovery resilience following a reduction in external phosphorus loading of shallow, eutrophic Danish lakes: duration, regulating factors and methods for overcoming resilience. Mem Ist Ital Idrobiol 48:127-148.
Jeppesen E, Søndergaard M, Mazzeo N, Meerhoff M, Branco CC, Huszar V and Scasso F (2005). Lake restoration and biomanipulation in temperate lakes: relevance for subtropical and tropical lakes. Restoration and Management of Tropical Eutrophic Lakes 341-359.
138 Jeppesen E et al. (2005). Lake responses to reduced nutrient loading - an analysis of contemporary long-term data from 35 case studies. Freshwater Biology 50:1747-1771.
Jöhnk KD, Huisman J, Sharples J, Sommeijer B, Visser PM, Stroom JM (2008). Summer heatwaves promote blooms of harmful cyanobacteria. Global Change Biology 14:495-512.
K
Kaplan-Levy RN, Hadas O, Summers ML, Rücker J, Sukenik A (2010). Akinetes: dormant cells of cyanobacteria In: Dormancy and resistance in harsh environments Springer, Berlin, Heidelberg. 5-27.
Kincaid TM, Olsen AR (2011) spsurvey: spatial survey design and analysis. R package version 2.5.
Kinsman R, Ibelings BW, Walsby A (1991). Gas vesicle collapse by turgor pressure and its role in buoyancy regulation by Anabaena flos-aquae. Journal of general microbiology 137:1171-1178.
Klausmeier CA, Litchman E, Daufresne T, Levin SA (2004). Optimal nitrogen-to- phosphorus stoichiometry of phytoplankton. Nature 429: 171.
Koenings JP, Edmundson JA (1991). Secchi disk and photometer estimates of light regimes in Alaskan lakes: Effects of yellow color and turbidity. Limnology and Oceanography 36(1):91-105.
Kohler SJ, Kothawala D, Futter MN, Liungman O, Tranvik L (2013). In-lake processes offset increased terrestrial inputs of dissolved organic carbon and colour to lakes. PloS one 8(8): e70598.
Kokociński M, Cameán AM, Carmeli S, Guzmán-Guillén R, Jos Á, Mankiewicz- Boczek J, Metcalf JS, Moreno IM, Prieto AI, Sukenik A. (2017). Cylindrospermopsin
139 and Congeners. In Meriluoto J, Spoof L, Codd GA (eds) Handbook of cyanobacterial monitoring and cyanotoxin analysis. John Wiley & Sons, Ltd.
Komárek J, Mareš J (2012). An update to modern taxonomy (2011) of freshwater planktic heterocytous cyanobacteria. Hydrobiologia 698:327-351.
Köppen, W(1900). Versuch einer Klassifikation der Klimate, vorzugsweise nach ihren Beziehungen zur Pflanzenwelt. GeogrZeitschr 6:593-679.
Kosten S et al. (2012). Warmer climates boost cyanobacterial dominance in shallow lakes. Global Change Biology 18:118-126.
Kruk C et al. (2010). A morphological classification capturing functional variation in phytoplankton. Freshwater Biology 55:614-627.
Kruk C et al. (2011). Phytoplankton community composition can be predicted best in terms of morphological groups. Limnology and Oceanography 56:110–118.
Kulkarni S, Deo MC, Ghosh S (2014). Changes in the design and operational wind due to climate change at the Indian offshore sites. Marine Structures 37:33-53.
Kurmayer R, Dittmann E, Fastner J, Chorus I (2002). Diversity of microcystin genes within a population of the toxic cyanobacterium Microcystis spp. in Lake Wannsee (Berlin, Germany). Microbial ecology 43:107-118.
Kurmayer R, Sivonen K, Salmaso N (2017). Introduction. In: Kurmayer R, Sivonen K, Wilmotte A, Salmaso, N (eds) Molecular tools for the detection and quantification of toxigenic cyanobacteria. John Wiley & Sons Ltd.
L
Lammens EH, van Nes EH, Meijer M-L, van den Berg MS (2004). Effects of commercial fishery on the bream population and the expansion of Chara aspera in Lake Veluwe. Ecological Modelling 177:233-244.
140 Lapierre JF and Giorgio PA (2012). Geographical and environmental drivers of regional differences in the lake pCO2 versus DOC relationship across northern landscapes. Journal of Geophysical Research: Biogeosciences 117.
Leach TH, Beisner BE, Carey CC, Pernica P, Rose KC, Huot Y, Verburg P (2017). Patterns and drivers of deep chlorophyll maxima structure in 100 lakes: The relative importance of light and thermal stratification. Limnology and Oceanography 63(2):628-646.
Legendre P, Legendre L (2012). Numerical ecology.Third English edition. Elsevier, Amsterdam, Netherlands.
Legendre P, Gallagher ED (2001). Ecologically meaningful transformations for ordination of species data. Oecologia 29:271-280.
Levine S, Schindler D (1999). Influence of nitrogen to phosphorus supply ratios and physicochemical conditions on cyanobacteria and phytoplankton species composition in the Experimental Lakes Area Canada. Canadian Journal of Fisheries and Aquatic Sciences 56:451-466.
Lewis Jr. WM , Wurtsbaugh WA (2008). Control of lacustrine phytoplankton by nutrients: erosion of the phosphorus paradigm. International Review of Hydrobiology 93(4-5):446-465.
Li R et al. (2001). Isolation and identification of the cyanotoxin cylindrospermopsin and deoxy-cylindrospermopsin from a Thailand strain of Cylindrospermopsis raciborskii (Cyanobacteria). Toxicon 39:973-980.
Li WKW, Carmack EC, McLaughlin FA, Nelson RJ, Williams WJ (2013). Space-for- time substitution in predicting the state of picoplankton and nanoplankton in a changing Arctic Ocean. Journal of Geophysical Research: Oceans 118:5750-5759.
141 Liere LV, Mur LR (1979). Growth kinetics of Oscillatoria agardhii Gomont in continuous culture, limited in its growth by the light energy supply. Journal of General Microbiology 115:153-160.
Liere LV, Walsby A (1982). Interactions of cyanobacteria with light. Botanical monographs 19:9-45.
Litchman E, Klausmeier CA (2008). Trait-based community ecology of phytoplankton. Annual Review of Ecology Evolution and Systematics 39:615-639.
Litchman E, Pinto PD, Klausmeier CA, Thomas MK, Yoshiyama K (2010). Linking traits to species diversity and community structure in phytoplankton. Hydrobiologia 653:15-28.
Livingstone DM, Lotter AF (1998). The relationship between air and water temperatures in lakes of the Swiss Plateau: a case study with paleolimnological implications. Journal of Paleolimnology 19(2):181-198.
Livingstone DM (2003). Impact of secular climate change on the thermal structure of a large temperate central European lake. Climatic Change 57:205-225.
Loftin KA, Graham JL, Hilborn E, Lehmann S, Meyer MT, Dietze JE, Griffith C (2016). Cyanotoxins in inland lakes of the United States: Occurrence and potential recreational health risks in the EPA National Lakes Assessment. Harmful Algae 56:77-90.
Loftin KA, Dietze JE, Meyer MT, Graham JL, Maksimowicz MM, Toyne KD (2016). Total cylindrospermopsins, microcystins/nodularins, and saxitoxins data for the 2007 United States Envrionmnetal Protection Agency National Lake Assessment. U.S. Geological Survey Data Series 929:9.
Long BM, Jones GJ, Orr PT (2001). Cellular microcystin content in N-Limited Microcystis aeruginosa can be predicted from growth rate. Applied and Environmental Microbiology 67(1):278-283.
142 Lürling M, Faassen EJ (2013). Dog poisonings associated with a Microcystis aeruginosa bloom in the Netherlands. Toxins 5:556-567.
Lürling M, Eshetu F, Faassen E, Kosten S, Huszar VL (2013). Comparison of cyanobacterial and green algal growth rates at different temperatures. Freshwater Biology 58:552-559.
Lürling M, van Oosterhout F, Faassen E (2017). Eutrophication and warming boost cyanobacterial biomass and microcystins. Toxins 9:64.
M
Mantzouki E, Visser PM, Bormans M, Ibelings BW (2016). Understanding the key ecological traits of cyanobacteria as a basis for their management and control in changing lakes. Aquatic Ecology 50:333-350.
Mantzouki E et al. (2018). Temperature effects explain continental scale distribution of cyanobacterial toxins. Toxins 10(4):156.
Mantzouki E, Ibelings BW (2018). The Principle and Value of the European Multi Lake Survey. Limnology and Oceanography Bulletin.
Marce R et al. (2016) Automatic High Frequency Monitoring for improved lake and reservoir management. Environmental Science and Technology 50:10780-10794.
Margalef R (1978). Life forms of phytoplankton as survival alternatives in an unstable environment. Oceanologica Acta 1:493-509.
Martín EJ, Doering M, Robinson CT (2017). Ecological assessment of a sediment by- pass tunnel on a receiving stream in Switzerland. River Research and Applications 33(6):925-936.
143 Meriluoto J et al. (2017). Toxic cyanobacteria and cyanotoxins in European waters – recent progress achieved through the CYANOCOST Action and challenges for further research. AIOL 8.
Messineo V, Melchiorre S, Di Corcia A, Gallo P, Bruno M (2010). Seasonal succession of Cylindrospermopsis raciborskii and Aphanizomenon ovalisporum blooms with cylindrospermopsin occurrence in the volcanic Lake Albano, Central Italy. Environmental Toxicology 25(1):18-27.
Milstead WB, Hollister JW, Moore RB, Walker HA (2013). Estimating summer nutrient concentrations in Northeastern lakes from SPARROW load predictions and modelled lake depth and volume. PloS one 8:e81457.
Mischke U (2003). Cyanobacteria associations in shallow polytrophic lakes: influence of environmental factors. Acta Oecologica 24:11-23.
Moe SJ, Dudley B, Ptacnik R (2008). REBECCA databases: experiences from compilation and analyses of monitoring data from 5,000 lakes in 20 European countries. Aquatic Ecology 42:183-201.
Mooij WM, Janse JH, deSenrpont Domis LN, Hülsmann S, Ibelings BW (2007). Predicting the effect of climate change on temperate shallow lakes with the ecosystem model PCLake. Hydrobiologia 584(1):443-454.
Monchamp ME, Walser JC, Pomati F, Spaak P (2016). Sedimentary DNA reveals cyanobacteria community diversity over 200 years in two peri-alpine lakes. Applied and environmental microbiology: AEM-02174.
Moss B et al. (2011). Allied attack: climate change and eutrophication. Inland Waters 1:101-105.
Mur L, Gons H, Liere L (1977). Some experiments on the competition between green algae and blue-green bacteria in light-limited environments. FEMS Microbiology Letters 1:335-338.
144 N
Neilan BA, Pearson LA, Muenchhoff J, Moffit MC, Dittmann E (2013). Environmental conditions that influence toxin biosynthesis in cyanobacteria. Environmental Microbiology 15(5):1239-53.
NETLAKE COST Action ES 1201: Networking Lake Observatories in Europe – https://www.dkit.ie/ga/node/14015 (2016).
NEN Water (1986). Photometric determination of the content of dissolved orthophosphate and the total content of phosphorous compounds by continuous flow analysis. Report NEN 6663 (Netherlands Normalization Institute).
NEN Water (1990). Photometric determination of the content of ammonium nitrogen and the sum of the contents of ammoniacal and organically bound nitrogen according to Kjeldahl by continuous flow analysis. Report NEN 6646 (Netherlands Normalization Institute).
NEN (1997). Bepaling van het stikstofgehalte in de vorm van nitriet en in de vorm vannitraat en de som van beide met doorstroomanalyse (CFA en FIA) en spectrometrische detectie. Report NEN-EN-ISO 13395 (Netherlands Normalization Institute).
Nimptsch J, Woelfl S, Osorio S, Valenzuela J, Moreira C, Ramos V, Castelo-Branco R, Leão PN, Vasconcelos V (2016). First record of toxins associated with cyanobacterial blooms in oligotrophic North Patagonian lakes of Chile- a genomic approach. International Review of Hydrobiology 101:57-68.
Nixdorf B, Mischke U, Rücker J (2003). Phytoplankton assemblages and steady state in deep and shallow eutrophic lakes—an approach to differentiate the habitat properties of Oscillatoriales. In: Phytoplankton and Equilibrium Concept: The Ecology of Steady-State Assemblages. Springer 111-121.
145 Noges P, Poikane S, Koiv T, Noges T (2010). Effect of chlorophyll sampling design on water quality assessment in thermally stratified lakes. Hydrobiologia 649:157-170.
Nogueira IC, Saker ML, Pflugmacher S, Wiegand C, Vasconcelos V (2004). Toxicity of the cyanobacterium Cylindrospermopsis raciborskii to Daphnia magna. Environmental Toxicology 19(5):453-9.
North RP, North RL, Livingstone DM, Köster O, Kipfer R (2014). Long-term changes in hypoxia and soluble reactive phosphorus in the hypolimnion of a large temperate lake: consequences of a climate regime shift. Global change biology 20:811-823.
Nydahl AC, Wallin MB and Weyhenmeyer GA (2017). No long-term trends in pCO2 despite increasing organic carbon concentrations in boreal lakes, streams, and rivers. Global Biogeochemical Cycles 31:985-995.
O
O’Brien KR, Meyer DL, Waite AM, Ivey GN, Hamilton DP (2004). Disaggregation of Microcystis aeruginosa colonies under turbulent mixing: laboratory experiments in a grid-stirred tank. Hydrobiologia 519:143-152.
Oksanen J, Kindt R, Legendre P, O'Hara B, Henry M, Stevens H (2007). The vegan package. Community ecology package 10: 631-637.
Otten TG, Xu H, Qin B, Zhu G, Paerl H.W (2012). Spatiotemporal patterns and ecophysiology of toxigenic Microcystis blooms in Lake Taihu, China: Implications for water quality management. Environmental Science & Technology 46:3480-3488.
P
Padisák J (1997). Cylindrospermopsis raciborskii (Woloszynska) Seenayya et Subba Raju, an expanding, highly adaptive cyanobacterium: worldwide distribution and review of its ecology. Archiv Für Hydrobiologie Supplementband Monographische Beitrage 107:563-593.
146 Paerl HW, Huisman J (2008). Blooms like it hot. Science 320(5872): 57-58.
Paerl HW, Huisman J (2009). Climate change: a catalyst for global expansion of harmful cyanobacterial blooms. Environmental Microbiology Reports 1:27-37.
Paerl HW, Hall NS, Calandrino ES (2011). Controlling harmful cyanobacterial blooms in a world experiencing anthropogenic and climatic-induced change. Science of the Total Environment 409:1739-1745.
Paerl HW, Hall NS, Calandrino ES (2011). Controlling harmful cyanobacterial blooms in a world experiencing anthropogenic and climatic-induced change. Science of the Total Environment 409(10):1739-1745.
Paerl HW, Otten TG (2013). Harmful cyanobacterial blooms: causes, consequences and controls. Microbial Ecology 65:995-1010.
Paerl HW, Otten TG (2013). Blooms bite the hand that feeds them. Science 342(6157): 433-434.
Paerl HW (2014). Mitigating harmful cyanobacterial blooms in a human- and climatically-impacted world. Life 4(4):988-1012.
Paerl HW et al. (2016). It takes two to tango: when and where dual nutrient (N & P) reductions are needed to protect lakes and downstream ecosystems. Environmental Science and Technology 50(20): 10805-10813.
Pawlik-Skowrońska B, Kalinowska R, Skowroński T (2013). Cyanotoxin diversity and food web bioaccumulation in a reservoir with decreasing phosphorus concentrations and perennial cyanobacterial blooms. Harmful Algae. 28:118-125.
Pickett STA (1989). Space-for-time substitution as an alternative to long-term studies. In: Long-term studies in ecology: approaches and alternatives. Springer, New York 110-135.
147 Pires LM, Sarpe D, Brehm M, Ibelings BW (2011). Potential synergistic effects of microcystins and bacterial lipopolysaccharides on life history traits of Daphnia galeata raised on low and high food levels. Aquatic Toxicology 104:230-42.
Pollard AI, Hampton SE, Leech DM (2018). The promise and potential of continental- scale limnology using the U.S Environmental Protection Agency's National Lakes Assessment. Limnology and Oceanography Bulletin 27(2): 36-41.
Pomati F, Jokela J, Simona M, Veronesi M, Ibelings BW (2011). An automated platform for phytoplankton ecology and aquatic ecosystem monitoring. Environmental science & technology 45:9658-9665.
Pomati F, Matthews B, Jokela J, Schildknecht A, Ibelings BW (2012). Effects of re- oligotrophication and climate warming on plankton richness and community stability in a deep mesotrophic lake. Oikos 121(8): 1317-1327.
Pomati F, Tellenbach C, Matthews B, Venail P, Ibelings BW, Ptacnik R (2015). Challenges and prospects for interpreting long-term phytoplankton diversity changes in Lake Zurich (Switzerland). Freshwater biology 60:1052-1059.
Portielje R, Van der Molen DT (1999). Relationships between eutrophication variables: from nutrient loading to transparency. In Shallow Lakes, Springer pp. 375- 387.
Posch T, Koester O, Salcher MM, Pernthaler J (2012). Harmful filamentous cyanobacteria favoured by reduced water turnover with lake warming. Nature Climate Change 2:809-813.
Ptacnik R, Solimin AG, Brettum P (2009). Performance of a new phytoplankton composition metric along a eutrophication gradient in Nordic lakes. Hydrobiologia 633(1):75-82.
148 R
R Development Core Team (2016). A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria https://www.R- project.org/ .
Rantala A, Fewer DP, Hisbergues M, Rouhiainen L, Vaitomaa J, Börner T, Sivonen K (2003). Phylogenetic evidence for the early evolution of microcystin synthesis. PNAS 101(2):578-573.
Rapala J, Sivonen K, Luukkainen R, Niemelä SI (1993). Anatoxin-a concentration in Anabaena and Aphanizomenon under different environmental conditions and comparison of growth by toxic and non-toxic Anabaena-strains: A laboratory study. Journal of applied phycology 5:581-591.
Rapala J, Sivonen K (2008). Assessment of environmental conditions that favor hepatotoxic and neurotoxic Anabaena spp. strains cultured under light limitation and different temperatures. Microbial Ecology 36:181-192.
Reiss J, Bridle JR, Montoya JM, Woodward G (2009). Emerging horizons in biodiversity and ecosystem functioning research. Trends in Ecology & Evolution 24:505-514.
Reynolds CS (1984). Phytoplankton periodicity - the interactions of form, function and environmental variability. Freshwater Biology 14:111-142.
Reynolds CS (1997). Excellence in ecology: vegetation processes in the pelagic: a model for ecosystem theory. Ecology Institute, Germany.
Reynolds CS, Huszar V, Kruk C, Naselli-Flores L, Melo S (2002). Towards a functional classification of the freshwater phytoplankton. Journal of Plankton Research 24:417-428.
149 Reynolds, CS (2006). The ecology of phytoplankton. Cambridge, Cambridge University Press.
Rigosi A, Carey CC, Ibelings BW, Brookes JD (2014). The interaction between climate warming and eutrophication to promote cyanobacteria is dependent on trophic state and varies among taxa. Limnology and Oceanography 59(1): 99-114.
Rockstrom JJ, Steffen WW, Noone KK, Persson AA, Chapin FSA (2009). Planetary boundaries: exploring the safe operating space for humanity. Ecology and Society 14(2): 32.
Roozen FC et al. (2003). Lake age and water level affect the turbidity of floodplain lakes along the lower Rhine. Freshwater Biology 48(3): 519-531.
Rubel F, Kottek M (2010). Observed and projected climate shifts 1901-2100 depicted by world maps of the Köppen-Geiger climate classification. Meteorologische Zeitschrift 19(2):135-141.
Ruiz-Jaen MC, Aide TM (2005). Restoration success: how is it being measured? Restoration ecology 13:569-577.
Ryan EF, Hamilton DP, Barnes GP (2004). Recent occurrence of Cylindrospermopsis raciborskii in Waikato lakes of New Zealand. New Zealand Journal of Marine and Freshwater Research 37:829-836.
Rzymski P, Poniedzialek B (2014). In search of environmental role of cylindrospermopsin: a review on global distribution and ecology of its producers. Water Research 66:320-337.
S
Salmaso N (2002). Ecological patterns of phytoplankton assemblages in Lake Garda: seasonal, spatial and historical features. Journal of Limnology 61(1):95-115.
150 Salmaso N et al. (2013). Influence of atmospheric modes of variability on the limnological characteristics of large lakes south of the Alps: a new emerging paradigm. Hydrobiologia 731: 31-48.
Salmaso N, Copetti D, Cerasino L, Shams S, Capelli C, Boscaini A, Guzzella L (2014). Variability of microcystin cell quota in metapopulations of Planktothrix rubescens: causes and implications for water management. Toxicon 90:82-96.
Salmaso N, Cerasino L, Boscaini A, Capelli C (2016). Planktic Tychonema (Cyanobacteria) in the large lakes south of the Alps: phylogenetic assessment and toxigenic potential. FEMS Microbiology Ecology 92:10.
Sas H (1989). Lake restoration by reduction of nutrient loading: Expectations, experiences, extrapolations. Internationale Vereinigung für theoretische und angewandte Limnologie: Verhandlungen 24(1): 247-251.
Scheffer M, Rinaldi S, Gragnani A, Mur LR, van Nes EH (1997). On the dominance of filamentous cyanobacteria in shallow, turbid lakes. Ecology 78:272-282.
Scheffer M, van Nes EH (2007). Shallow lakes theory revisited: various alternative regimes driven by climate, nutrients, depth and lake size. In: Shallow lakes in a changing world, Springer pp 455-466.
Scheffer M et al. (2012). Anticipating critical transitions. Science 338:344-348.
Schindler D (1977). Evolution of phosphorus limitation in lakes. Science 195:260-262.
Schindler DW, Vallentyne JR (2008). The algal bowl, overfertilization of the world's freshwaters and estuaries. University of Alberta Press, Earthscan.
Schindler DW et al. (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.
151 Schindler DW, Carpenter SR, Chapra SC, Hecky RE, Orihel DM (2016). Reducing phosphorus to curb lake eutrophication is a success. Environmental Science and Technology 50: 8923–8929.
Schopf JW (2002) The fossil record: tracing the roots of the cyanobacterial lineage in the ecology of cyanobacteria: their diversity in time and space. In Whitton BA, Potts M (eds) Springer Netherlands 13-35.
Schwaderer AS, Yoshiyama K, Pinto PT, Swenson NG, Klausmeier CA, Litchman E (2011). Eco-evolutionary differences in light utilization traits and distributions of freshwater phytoplankton. Limnology and Oceanography 56: 589-598.
Schwefel R, Gaudard A, Wuest A, Bouffard D (2016). Effects of climate change on deepwater oxygen and winter mixing in a deep lake (Lake Geneva): Comparing observational findings and modeling. Water Resource Research 52:8811-8826.
Scott JT, McCarthy MJ (2010). Nitrogen fixation may not balance the nitrogen pool in lakes over timescales relevant to eutrophication management. Limnology and Oceanography 55:1265-1270.
Sevilla E, Martin-Luna B, Vela L, Bes MT, Peleato ML, Fillat MF (2010). Microcystin- LR synthesis as response to nitrogen: transcriptional analysis of the mcyD gene in Microcystis aeruginosa PCC7806. Ecotoxicology 19:1167–1173.
Shinohara R, Isobe M (2010). Daily change in wind-induced sediment resuspension and phosphorus forms in a shallow eutrophic lake. Fundamental and Applied Limnology 176:161-171.
Sinha R, Pearson LA, Davis TW, Burford MA, Orr PT, Neilan BA (2012). Increased incidence of Cylindrospermopsis raciborskii in temperate zones - Is climate change responsible? Water Research 46:1408-1419.
152 Sivonen K (1990). Effects of light, temperature, nitrate, orthophosphate, and bacteria on growth of and hepatotoxin production by Oscillatoria agardhii strains. Applied and Environmental Micorbiology 56(9): 2658-2666.
Sivonen K, Jones G (1999). Toxic cyanobacteria in water: a guide to their public health consequences, monitoring and management. (eds) Chorus I, Bertram J. E & FN Spon: London, UK, 41-111.
Skjelkvale BL, Henriksen A, Jonsson GS, Mannio J, Wilander A, Jensen JP, Fjeld E, Lien L (2001). Chemistry of lakes in the Nordic region - Denmark, Finland with Aland, Iceland, Norway with Svalbard and Bear Island, and Sweden (4391-2001 RS).
Smayda TJ (1970). The suspension and sinking of phytoplankton in the sea. Oceanography and Marine Biology Annual Review 8:353-414.
Solomon S (2007). Climate change 2007-the physical science basis. Working group I contribution to the fourth assessment report of the IPCC Vol4.Cambridge University Press.
Sommer U et al. (2012). Beyond the Plankton Ecology Group (PEG) model: mechanisms driving plankton succession. Annual Review of Ecology, Evolution and Systematics 43: 429-448.
Søndergaard M, Jeppesen E (2007). Anthropogenic impacts on lake and stream ecosystems, and approaches to restoration. Journal of applied ecology 44:1089-1094.
Soranno PA, Schimel DS (2014). Macrosystems ecology: big data, big ecology. Frontiers in Ecology and the Environment 12:3-3.
Soranno PA et al. (2015). Building a multi-scaled geospatial temporal ecology database from disparate data sources: fostering open science and data reuse. GigaScience 4(1): 28.
153 Soranno PA et al. (2017). LAGOS-NE: a multi-scaled geospatial and temporal database of lake ecological context and water quality for thousands of US lakes. GigaScience 6:1-22.
Sosik HM, Olson RJ (2007). Automated taxonomic classification of phytoplankton sampled with imaging-in-flow cytometry. Limnology and Oceanography: Methods 5:204-216.
Staal J, Hlobil H, Van Tulder M, Waddell G, Burton AK, Koes B, Van Mechelen W (2003). Occupational health guidelines for the management of low back pain: an international comparison. Occupational and environmental medicine 60:618-626.
Steffensen DA (2008). Economic cost of cyanobacterial blooms. In Cyanobacterial harmful algal blooms: State of the science and research needs (pp. 855-865). Springer, New York, NY.
Stevens Jr. DL, Olsen AR (2004). Spatially balanced sampling of natural resources. Journal of the American Statistical Association 99(465):262-278.
Stomp M, van Dijk MA, van Overzee HMJ, Wortel MT, Sigon CAM, Egas M, Hoogvelt H, Gons HJ, Huisman J (2008). The timescale of phenotypic plasticity and its impact on competition in fluctuating environments. American Naturalist 172(5): 169-185.
Straile D, Kerimoglu O, Peeters F, Jochimsen MC, Kümmerlin R, Rinke K, Rothhaupt KO (2010). Effects of a half a millennium winter on a deep lake–a shape of things to come? Global Change Biology 16:2844-2856.
Straile D, Jochimsen MC, Kümmerlin R (2013). The use of long-term monitoring data for studies of planktonic diversity: a cautionary tale from Swiss lakes. Freshwater Biology 58(6): 1292-1301.
154 Stüken A et al. (2006). Distribution of three alien cyanobacterial species (Nostocales) in northeast Germany: Cylindrospermopsis raciborskii, Anabaena bergii and Aphanizomenon aphanizomenoides. Phycologia 45:696-703.
Suding KN et al. (2008). Scaling environmental change through the community- level: a trait-based response-and-effect framework for plants. Global Change Biology 14:1125-1140.
T
Taranu ZE, Zurawell RW, Pick F, Gregory-Eaves I (2012). Predicting cyanobacterial dynamics in the face of global change: the importance of scale and environmental context. Global Change Biology 18:3477-3490.
Thomas MK, Kremer CT, Litchman E (2016). Environment and evolutionary history determine the global biogeography of phytoplankton temperature traits. Global Ecology and Biogeography 25(1):75-86.
Thomas MK, Litchman E (2016). Effects of temperature and nitrogen availability on the growth of invasive and native cyanobacteria. Hydrobiologia 763: 357-369.
Toming K, Kutser T, Laas A, Sepp M, Paavel B, Nõges T (2016) .First experiences in mapping lake water quality parameters with Sentinel-2 MSI imagery. Remote Sensing 8:640.
Tonk L et al. (2005). The microcystin composition of the cyanobacterium Planktothrix agardhii changes toward a more toxic variant with increasing light intensity. Applied and Environmental Microbiology 71(9): 5177-5181.
Tüxen R (1955). Das system der nordwestdeutschen Pflanzengesellschaften. Floristisch-soziolog Arbeitsgemeinschaft.
155 U
UNEP (2016). Loss and Damage: The role of Ecosystem Services. United Nations Environment Programme, Nairobi, Kenya.
V
Van der Staay GWM, Brouwer A, Baard RL, van Mourik F (1992). Separation of photosystems I and II from the oxychlorobacterium (prochlorophyte) Prochlorothrix hollandica and association of Chlb binding antennae with PS II. Biochimica et Biophysica Acta 1102:220-228.
Vaughn CC (2010). Biodiversity losses and ecosystem function in freshwaters: emerging conclusions and research directions. Bioscience 60:25-35.
Vautard R, et al. (2014). The European climate under a 2 °C global warming. Environmental Research Letters 9(3):034006.
Viney NR et al. (2007). Modelling adaptive management strategies for coping with the impacts of climate variability and change on riverine algal blooms. Global Change Biology 13:1-13.
Venables WN, Ripley BD (2002). Modern Applied Statistics with S-Plus. Fourth Edition Springer New York.
Verspagen JM et al. (2006). Water management strategies against toxic Microcystis blooms in the Dutch delta. Ecological applications 16:313-327.
Verspagen JM et al. (2014). Rising CO2 levels will intensify phytoplankton blooms in eutrophic and hypertrophic lakes. PLOS ONE 9(8): e104325.
Visser PM, Ibelings BW, Mur LR (1995). Autumnal sedimentation of Microcystis spp as result of an increase in carbohydrate ballast at reduced temperature. Journal of Plankton Research 17:919-933.
156 Visser P, Ibelings B, van der Veer B, Koedood J, Mur R (1996). Artificial mixing prevents nuisance blooms of the cyanobacterium Microcystis in Lake Nieuwe Meer, the Netherlands. Freshwater Biology 36:435-450.
Visser PM, Ibelings BW, Mur LR, Walsby AE (2005). The ecophysiology of the harmful cyanobacterium Microcystis In: Harmful Cyanobacteria Springer 109-142.
Visser PM, Ibelings BW, Bormans M, Huisman J (2016). Artificial mixing to control cyanobacterial blooms: a review. Aquatic Ecology 50(3):423-441.
Vitousek PM et al. (2002). Towards an ecological understanding of biological nitrogen fixation. Biogeochemistry 57:1-45.
Vos RJ, Hakvoort JHM, Jordans RWJ, Ibelings BW (2003). Multiplatform optical monitoring of eutrophication in temporally and spatially variable lakes. Science of the Total Environment 312:221-243.
Vrede T, Ballantyne A, Mille-Lindblom C, Algesten G, Gudasz C, Lindahl S, Brunberg AK (2009). Effects of N: P loading ratios on phytoplankton community composition, primary production and N fixation in a eutrophic lake. Freshwater Biology 54:331-344.
W
Wacklin P, Hoffmann L, Komárek J (2009). Nomenclatural validation of the genetically revised cyanobacterial genus Dolichospermum (Ralfs ex Bornet et Flahault) comb nova. Fottea 9(1): 59-64.
Wagner C, Adrian R (2009). Cyanobacteria dominance: Quantifying the effects of climate change. Limnology and Oceanography 54:2460-2468.
Wainwright CE, Staples TL, Charles LS, Flanagan TC, Lai HR, Loy X, Reynolds VA and Mayfield MM (2018). Links between community ecology theory and ecological restoration are on the rise. Journal of Applied Ecology 55:570-581.
157 Walsby AE (1991). The mechanical properties of the Microcystis gas vesicle. Journal of General Microbiology 137:2401-2408.
Walsby AE (1994). Gas vesicles. Microbiological reviews 58:94.
Walsby AE (2005). Stratification by cyanobacteria in lakes: a dynamic buoyancy model indicates size limitations met by Planktothrix rubescens filaments. New Phytologist 168:365-376.
Walsby AE, Avery A, Schanz F (1998). The critical pressures of gas vesicles in Planktorhrix rubescens in relation tothe depth of winter mixing in Lake Zürich, Switzerland. Journal of Plankton Research 20:1357-1375.
Walsby AE, Ng G, Dunn C, Davis PA (2004). Comparison of the depth where Planktothrix rubescens stratifies and the depth where the daily insolation supports its neutral buoyancy. New Phytologist 162:133-145.
Weyhenmeyer GA, Meili M and Livingstone DM (2004). Nonlinear temperature response of lake ice breakup. Geophysical Research Letters 31:7.
Weyhenmeyer GA (2009). Increasing dissimilarity of water chemical compositions in a warmer climate. Global Biogeochemical Cycles 23:2.
Weyhenmeyer GA, Jeppesen E(2009). Nitrogen deposition induced changes in DOC:NO3-N ratios determine the efficiency of nitrate removal from freshwaters. Global Change Biology 16(8): 2358-2365.
Weyhenmeyer GA, Karlsson J (2009). Nonlinear response of dissolved organic carbon concentrations in boreal lakes to increasing temperatures. Limnology and Oceanography 54:2513-2519.
Weyhenmeyer GA, Mackay M, Stockwell JD, Thiery W, Grossart HP, Augusto-Silva PB, Baulch HM, de Eyto E, Hejzlar J, Kangur K, Kirillin G, Pierson DC, Rusak JA, Sadro S and Woolway RI (2017). Citizen science shows systematic changes in the
158 temperature difference between air and inland waters with global warming. Scientific Reports 7:43890.
Wiedner C, Visser PM, Fastner J, Metcalf JS, Codd GA, Mur LR (2003). Effects of light on the microcystin content of Microcystis strain PCC 7806. Applied and Environmental Microbiology 69(3):1475-1481.
Williamson CE, Brentrup JA, Zhang J, Renwick WH, Hargreaves BR, Knoll LB, Rose KC (2014). Lakes as sensors in the landscape: Optical metrics as scalable sentinel responses to climate change. Limnology and Oceanography 59(3): 840-850.
Wilhelm S, Adrian R (2008). Impact of summer warming on the thermal characteristics of a polymictic lake and consequences for oxygen, nutrients and phytoplankton. Freshwater Biology 53(2): 226-237.
Williamson CE, Dodds W, Kratz TK, Palmer MA (2008). Lakes and streams as sentinels of environmental change in terrestrial and atmospheric processes. Frontiers in Ecology and the Environment 6(5):247-254.
Winslow LA, Read JS, Woolway RI, Brentrup JA, Leach TH, Zwart JA (2016) rLakeAnalyzer: Lake physics tools. Zenodo: https://zenodo.org/record/1003169#.WzYwh9Uza70 .
WHO (1996). Guidelines for drinking water quality-health criteria and other supporting information. World Health Organization, Geneva.
Woodward G (2009). Biodiversity, ecosystem functioning and food webs in fresh waters: assembling the jigsaw puzzle. Freshwater Biology 54:2171-2187.
Wood SA, Maier MY, Puddick J, Pochon X, Zaiko A, Dietrich DR, Hamilton DP (2016). Trophic state and geographic gradients influence planktonic cyanobacterial diversity and distribution in New Zealand lakes. FEMS Microbiol Ecology 93(2).
159 Wolf HU, Frank C (2002). Toxicity assessment of cyanobacterial toxin mixtures. Environmental Toxicology 17(4):395-9.
Y
Yankova Y, Neuenschwander S, Koster O, Posch T (2017). Abrupt stop of deep water turnover with lake warming: Drastic consequences for algal primary producers. Scientific Reports 7(1): 13770.
Yılmaz M, Phlips EJ, Szabo NJ, Badylak S (2008). A comparative study of Florida strains of Cylindrospermopsis and Aphanizomenon for cylindrospermopsin production. Toxicon 51(1): 130-139.
Young GM (2012) Precambrian supercontinents, glaciations, atmospheric oxygenation, metazoan evolution and an impact that may have changed the second half of Earth history. Geoscience Frontiers 4: 247-261.
160 Remerciements
Ce projet ne serait pas possible s'il n'y avait pas 200 scientifiques enthousiastes qui soutiennent la science collaborative. Plus important encore, cette thèse aurait été impossible si mon supérieur hiérarchique ne m'avait pas implicitement confié la tâche prodigieuse de coordonner des personnes de 27 pays. Ce fut un honneur de relever un tel défi et de le faire fonctionner de la meilleure façon possible. Je voudrais remercier Bas pour son soutien illimité, pour sa disponibilité à échanger des idées et pour avoir attendu patiemment que ce projet se concrétise. Son approche en tant que superviseur m'a permis d'apprendre beaucoup, de devenir plus sage, plus indépendant et plus confiant dans mes compétences en tant que scientifique. Je ne pouvais pas être plus chanceux.
Ce projet de doctorat m'a été présenté par la Dre Petra Visser, la personne qui a toujours le bon conseil pour moi. Elle savait que ce projet conviendrait parfaitement et a proposé de me soutenir en tant que personne de référence si je devais en faire la demande. Je voudrais la remercier pour sa disponibilité constante à m'écouter, même si elle ne travaille pas directement avec elle.
J'aimerais surtout remercier mes parents de m'avoir élevé dans un environnement parfait pour qu'un enfant développe son autosuffisance et son empathie. Bien que maman et papa aimeraient que je sois dans le même pays qu'eux, ils appuient totalement ma décision et s'assurent que je sache qu'ils sont fiers de moi. Je voudrais également remercier mes sœurs d’être les meilleures partenaires en matière de bêtise et d’insouciance.
Je voudrais remercier ma première étudiante en Master, Clare Ahnlund, qui a énormément aidé à diriger le Swiss Multi Lake Survey et à être patiente avec moi alors que j'apprenais à devenir superviseur! Dan, Daphne et Dominic, l'équipe de physique aquatique, merci beaucoup pour toutes ces heures consacrées à la réflexion sur des idées sur la façon de traiter le jeu de données EMLS et à la possibilité de répondre à tout type de question. Vos précieux commentaires ont accru ma compréhension de la physique aquatique, mais aussi des approches écologiques en général. Patrick, Irene, Zi, Xu, Fabio, Joren, Jorrit, Julio, Roxane, Cezar, Alonso, oh-Dieu-je-espère-je-n'oublie pas quelqu'un, merci pour les grandes discussions scientifiques et les bons moments
161 passés à Forel. Tonya, c’est un plaisir de vous accueillir ici et, bien que nous n’ayons pas encore travaillé ensemble, j’espère que nous le ferons à l’avenir, alors merci de revenir en Suisse! Alexandra, merci pour tous vos efforts et vos conseils pour faire face à la lourde bureaucratie et à toutes les discussions intéressantes sur les randonnées dans la nature. Katia, merci beaucoup d'avoir rendu nos besoins informatiques possibles, sans lesquels cette thèse ne serait pas possible.
Enfin et surtout des Foréliens, je voudrais remercier mes deux meilleures amies, Ena et Timon, pour le plaisir qu'elles ont eu au bureau, au laboratoire, sur le terrain, aux conférences, aux soirées, à la montagne, dans la vie avant tout! C’est formidable d’avoir des personnes sur lesquelles vous pouvez compter, en particulier lorsque l’on a besoin de se lamenter beaucoup sur son doctorat (juste pour le craic :-P).
Je ne saurais pas qui remercier d’abord des partenaires EMLS. De CyanoCOST: Jutta Fastner, Kristel Panksep, Nico Salmaso, Myriam Bormans, Hans Matthijs (peut reposer en paix), Tri Kaloudis, Theodore Triantis, Anastasia Hiskia, Spiros Gkelis, Maria Antoniou, Ludek Blaha, Ludek Blaha, Rainer Kurmayer, Svetislav Krstik , Trine Perlt Warming, Kirsten Christoferrsen, Judita Koreiviene, Iwona Jasser, Hanna Mazur-Marzec, Tina Elersek, Meric Albay, Reyhan Akçaalan, Sigrid Haande, Vitor Vasconcelos, Rafael Marche et bien d'autres personnes ayant souscrit à l'idée d'EML, sont devenues représentantes de leurs pays et ont aidé à attirer plus de contributeurs de données pour le succès complet de l’enquête. De même, de NETLAKE: Eleanor Jennings, Lauri Arvola, Elvira de Eyto, Julita Dunalska, Nusret Karakaya, Giovanna Flaim, Valerie McCarthy et bien d’autres, qui ont contribué à l’extension du système EML. Je voudrais remercier Elzbieta Wilk-Wozniak, Magdalena Toporowska, Justyna Kobos, Meryem Beklioglu, Ana Garcia, Carmen Cillero Castro et Joao Morais pour l'organisation d'écoles de formation et la coordination des plus grandes enquêtes nationales EMLS dans leurs pays respectifs (Pologne, Turquie, Espagne et Portugal).
Je tiens à remercier Lisette deSenerpont Domis pour son soutien dans le développement de l’objectif de l’EMLS, pour l’enseignement à l’école de formation EMLS, pour son soutien dans les analyses statistiques et la rédaction de manuscrits (avec Sven Teurlincx et Laura Seelen). Hans-Peter Grossart, merci d'être toujours le premier à nous faire part de vos commentaires sur tous les manuscrits que nous avons réunis, de votre soutien et de votre disponibilité pour des discussions. Cayelan Carey,
162 merci pour tout ce que vous avez fait pour l'EMLS sans être européen. C'est le véritable esprit d'un grand scientifique, sans frontières, mais avec de nombreuses idées intéressantes et un enthousiasme débordant. J'espère qu'un jour nous pourrons travailler plus étroitement. Mike Lurling, merci de croire en l’EMLS plus que quiconque (du moins au début!) Et de vous offrir un espace pour stocker et trier les échantillons EMLS, pour couvrir les coûts de nombreuses analyses et pour avoir invité notre premier article EMLS à le journal des toxines. Vos idées et votre efficacité ont été un catalyseur du succès européen.
Enfin et surtout, je voudrais remercier ma femme car elle est toujours à côté de moi malgré toutes les difficultés et leur donne un aspect facile. Merci pour votre soutien, votre patience, vos idées créatives et vos questions sans fin et curieuses! Merci pour la relecture de mes manuscrits et pour m'avoir parlé en anglais (donc peut-être qu'un jour vous n'aurez pas à relire mon travail!). Merci de me donner envie d'être une personne meilleure et de contribuer ensemble à un monde meilleur.
163 Acknowledgements
This project would never have been possible, had it not been for a couple of hundred excited scientists that support collaborative science. But to start with, this thesis would have been impossible, if my supervisor hadn’t implicitly trusted me with the prodigious task of coordinating people from 27 countries. Or maybe he didn’t trust me at first, but took the risk anyway! No matter what came first, it was an honor to take such a challenge and make it work the best way possible. I would like to thank Bas for being a great Doktorvater, for being available to provide his limitless support, to brainstorm on new ideas or solutions, and for waiting patiently for this project to come together. His approach as a supervisor has helped me learn a lot, grow wiser, more independent and more confident about my skills as a scientist. I could not be luckier. Thank you Bas, for also introducing me to the great GLEON crowds and for showing me how to drop the beat at the parties!
This doctoral project was made known to me by Dr. Petra Visser, my MSc’ mentor, the person that always has the right advice for me. She knew that this project would be a perfect fit, and offered to support me as a reference person if I were to apply for it. I would like to thank her warmly for this and for always being open to listen to me, despite not working with her directly.
I would like to thank most-importantly my parents for raising me in a perfect environment (not too spoiled, but with a lot of love and security) for a kid to develop self-sufficiency and empathy. Although mama and papa would love me to be in the same country as them, building a home close to theirs, they support my decision completely and they make sure I know they are proud of me. I would like to thank my sisters as well, for being the best partners in silliness and carefreeness!
Coming back to the Forel people, I would like to thank my first Master student, Clare Ahnlund that helped massively with running the Swiss Multi Lake Survey, and for being patient with me while I was learning how to be a supervisor! Dan, Daphne and Dominic, the Aquatic Physics team, thank you very much for all these hours you devoted in meeting with me, brainstorming ideas on how to deal with the EMLS dataset and for being around answering any type of question. Your invaluable feedback increased my understanding of not just aquatic physics, but ecological
164 approaches in general. Patrick, Irene, Zi, Xu, Fabio, Joren, Jorrit, Julio, Roxane, Cezar, Alonso, oh-God-I-hope-I-m-not-forgetting-someone, thank you for the great scientific – or not – discussions and good times around Forel. Tonya, it is great to have you here, and although we haven’t worked together yet, I hope we will in the future, so thank you for coming back to Switzerland! Alexandra, thank you for all your efforts and advice on dealing with the massive bureaucracy and all the interesting talks about rambling in nature. Katia, thank you very much for making our IT needs possible, without which this dissertation would be ugly.
Last but not least from the Forelians, I would like to thank my two best mates, Ena and Timon, for all the great fun in the office, in the lab, in the field, at the conferences, at my bachelorette, at the parties, in the mountains, in life over all! It is great to have people you can lean on, especially when one needs to whine a lot about one’s PhD (just for the craic :-P).
I would not know who to thank first from the EMLS partners. From CyanoCOST: Jutta Fastner, Kristel Panksep, Nico Salmaso, Myriam Bormans, Hans Matthijs (may rest in peace), Tri Kaloudis, Theodore Triantis, Anastasia Hiskia, Spiros Gkelis, Maria Antoniou, Ludek Blaha, Rainer Kurmayer, Svetislav Krstik, Andrea Torokne, Trine Perlt Warming, Kirsten Christoferrsen Judita Koreiviene, Iwona Jasser, Hanna Mazur- Marzec, Tina Elersek, Meric Albay, Reyhan Akçaalan, Sigrid Haande, Vitor Vasconcelos, Rafael Marche and many more, that supported the idea of the EMLS, became the representative of their countries and helped in attracting more data contributors for the complete success of the survey. Similarly, from NETLAKE: Eleanor Jennings, Lauri Arvola, Elvira de Eyto, Julita Dunalska, Nusret Karakaya, Giovanna Flaim, Valerie McCarthy and many more, that helped to expand the EMLS. I would like to thank Elzbieta Wilk-Wozniak, Magdalena Toporowska, Justyna Kobos, Meryem Beklioglu, Ana Garcia, Carmen Cillero Castro and Joao Morais for organising training schools and coordinating the largest of the national EMLS surveys, in their respective countries (Poland, Turkey, Spain & Portugal).
I would like to thank Lisette deSenerpont Domis, for all her support in developing the purpose of the EMLS, in teaching the EMLS training school, in supporting me with the statistical analyses and manuscript writing (along with Sven Teurlincx and Laura Seelen); but most importantly for all our talks that helped me embrace who I am and
165 build on it. Hans-Peter Grossart, thank you for always being the first to return comments on all the manuscripts we have together, for being supportive and available for discussions. Cayelan Carey, thank you for everything you have done for the EMLS despite not being European. This is the true spirit of a great scientist, with no borders but numerous interesting ideas and enthusiasm. I hope that one day we get to work closer. Mike Lurling, thank you for believing in the EMLS more than anyone else (at least in the beginning!) and for providing all the facilities and work effort in storing and sorting out the EMLS messy samples, for covering the costs of many analyses and for inviting our first EMLS paper to the special issue of Toxins. Your ideas and efficiency were a catalyst for European success.
Finally yet importantly, I would like to thank my wife because she is always next to me through all difficulties and makes them look easy. Thank you for all your support, all your patience, all your creative ideas and endless, curious questions! Thank you for proof reading my manuscripts and for talking to me in articulate English (so maybe one day you will not have to proof read my work!). Thank you for making me want to be a better person and contribute together to a better world.
166 Authors and Affiliations Chapter 3, 4 and 5 were the three predecided papers that set the aim of the European Multi Lake Survey (EMLS). During the training school in Evian-Les-Bains (May 2015), we agreed on including all data contributors to the co-authors’ list of those EMLS papers.
Authors
Evanthia Mantzouki 1, *, James Campbell2, Emiel van Loon3, Petra Visser3, Iosif Konstantinou4, Maria Antoniou4, Grégory Giuliani5, Danielle Machado-Vieira6, Alinne Gurjão de Oliveira6, Dubravka Špoljarić Maronić7, Filip Stević7, Tanja Žuna Pfeiffer7, Itana Bokan Vucelić8, Petar Žutinić9, Marija Gligora Udovič9, Anđelka Plenković- Moraj9, Nikoletta Tsiarta4, Luděk Bláha10, Rodan Geriš11, Markéta Fránková12, Kirsten Seestern Christoffersen13, Trine Perlt Warming13, Tõnu Feldmann14, Alo Laas14, Kristel Panksep14, Lea Tuvikene14, Kersti Kangro14, 15, Kerstin Häggqvist16, Pauliina Salmi17, Lauri Arvola18, Jutta Fastner19, Dietmar Straile20, Karl Rothhaupt20, Jeremy Fonvielle21, Hans-Peter Grossart21,22, Christos Avagianos23, Triantafyllos Kaloudis23, Theodoros Triantis24, Sevasti-Kiriaki Zervou24, Anastasia Hiskia24, Spyros Gkelis25, Manthos Panou25, Valerie McCarthy26, Victor C. Perello26, Ulrike Obertegger27, Adriano Boscaini27, Giovanna Flaim27, Nico Salmaso27, Leonardo Cerasino27, Judita Koreivienė28, Jūratė Karosienė28, Jūratė Kasperovičienė28, Ksenija Savadova28, Irma Vitonytė28, Sigrid Haande29, Birger Skjelbred29, Magdalena Grabowska30, Maciej Karpowicz30, Damian Chmura31, Lidia Nawrocka32, Justyna Kobos33, Hanna Mazur- Marzec33, Pablo Alcaraz-Párraga34, Elżbieta Wilk-Woźniak35, Wojciech Krztoń35, Edward Walusiak35, Ilona Gagala36, Joana Mankiewicz-Boczek36, Magdalena Toporowska37, Barbara Pawlik-Skowronska37, Michał Niedźwiecki37, Wojciech Pęczuła37, Agnieszka Napiórkowska-Krzebietke38, Julita Dunalska39, Justyna Sieńska39, Daniel Szymański39, Marek Kruk40, Agnieszka Budzyńska41, Ryszard Goldyn41, Anna Kozak41, Joanna Rosińska41, Elżbieta Szeląg-Wasielewska41, Piotr Domek41, Natalia Jakubowska-Krepska41, Kinga Kwasizur42, Beata Messyasz42, Aleksandra Pełechata42, Mariusz Pełechaty42, Mikolaj Kokocinski42, Beata Madrecka43, Iwona Kostrzewska- Szlakowska44, Magdalena Frąk45, Agnieszka Bańkowska-Sobczak46, Michał Wasilewicz46, Agnieszka Ochocka47, Agnieszka Pasztaleniec47, Iwona Jasser48, Ana M. Antão-Geraldes49, Manel Leira50, Armand Hernández51, Vitor Vasconcelos52, Joao
167 Morais52, Micaela Vale52, Pedro M. Raposeiro53, Vítor Gonçalves53, Boris Aleksovski54, Svetislav Krstić54, Hana Nemova55, Iveta Drastichova55, Lucia Chomova55, Spela Remec-Rekar56, Tina Elersek57, Jordi Delgado-Martín58, David García58, Jose Luís Cereijo58, Joan Gomà59, Mari Carmen Trapote59, Teresa Vegas-Vilarrúbia59, Biel Obrador59, Ana García-Murcia60, Monserrat Real60, Elvira Romans60, Jordi Noguero- Ribes60, David Parreño Duque60, Elísabeth Fernández-Morán60, Bárbara Úbeda61, José Ángel Gálvez61, Rafael Marcé62, Núria Catalán62, Carmen Pérez-Martínez63, Eloísa Ramos-Rodríguez63, Carmen Cillero-Castro64, Enrique Moreno-Ostos65, José María Blanco65, Valeriano Rodríguez65, Jorge Juan Montes-Pérez65, Roberto L. Palomino65, Estela Rodríguez-Pérez65, Rafael Carballeira66, Antonio Camacho67, Antonio Picazo67, Carlos Rochera67, Anna C. Santamans67, Carmen Ferriol67, Susana Romo68, Juan Miguel Soria68, Lars-Anders Hansson69, Pablo Urrutia-Cordero69,71, Arda Özen70, Andrea G. Bravo71, Moritz Buck71, William Colom-Montero72, Kristiina Mustonen72, Don Pierson72, Yang Yang72, Jolanda Verspagen3, Lisette N. de Senerpont Domis73,74, Laura Seelen73,74, Sven Teurlincx73, Yvon Verstijnen74, Miquel Lürling73,75, Valentini Maliaka74,76,77, Elisabeth J. Faassen74,78, Delphine Latour79, Cayelan C. Carey80, Hans Paerl81, Andrea Torokne82, Tünay Karan83, Nilsun Demir84, Meryem Beklioğlu85, Nur Filiz85, Eti Levi85, Uğur Iskin85, Gizem Bezirci85, Ülkü Nihan Tavşanoğlu85, Kemal Çelik86, Koray Ozhan87, Nusret Karakaya88, Mehmet Ali Turan Koçer89, Mete Yilmaz90, Faruk Maraşlıoğlu91, Özden Fakioglu92, Elif Neyran Soylu93, Meral Apaydın Yağcı94, Şakir Çınar94, Kadir Çapkın94, Abdulkadir Yağcı94, Mehmet Cesur94, Fuat Bilgin94, Cafer Bulut94, Rahmi Uysal94, Köker Latife95, Reyhan Akçaalan95, Meriç Albay95, Mehmet Tahir Alp96, Korhan Özkan97, Tuğba Ongun Sevindik98, Hatice Tunca98, Burçin Önem98, Jessica Richardson99, Christine Edwards100, Victoria Bergkemper101, Sarah O'Leary102, Eilish Beirne103 and Bastiaan W. Ibelings1
Affiliations
1. Department F.-A. Forel for Environmental and Aquatic Sciences, University of Geneva, 1205 Geneva, Switzerland
2. Institute of Biology, Leiden University, 2333 BE Leiden, The Netherlands
3. Department of Freshwater and Marine Ecology, University of Amsterdam, 1090 GE Amsterdam, Netherlands
168 4. Department of Environmental Science and Technology, Cyprus University of Technology, 3036 Lemesos, Cyprus
5. Institute for Environmental Sciences - GRID, University of Geneva, 1211 Geneva, Switzerland
6. Departamento de Sistemática e Ecologia, Universidade Federal da Paraíba, 58059-970 Paraíba, Brasil
7. Department of Biology, Josip Juraj Strossmayer University of Osijek, 31000 Osijek, Croatia
8. Department for Ecotoxicology, Teaching Institute of Public Health of Primorje- Gorski Kotar County, 51000 Rijeka, Croatia
9. Department of Biology, Faculty of Science, University of Zagreb, 10000 Zagreb, Croatia
10. RECETOX, Faculty of Science, Masaryk University, 62500 Brno, Czech Republic
11. Department of Hydrobiology, Morava Board Authority, 60200 Brno, Czech Republic
12. Laboratory of Paleoecology, Institute of Botany, The Czech Academy of Sciences, 60200 Brno, Czech Republic
13. Freshwater Biological Laboratory, Department of Biology, University of Copenhagen, 2100 Copenhagen, Denmark
14. Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, 51014 Tartu, Estonia
15. Tartu Observatory, Faculty of Science and Technology, University of Tartu, 61602 Tartu, Estonia
16. Department of Science and Engineering, Åbo Akademi University, 20520 Åbo, Finland
17. Department of Biological and Environmental Science, University of Jyväskylä, 40014 Jyväskylä, Finland
18. Lammi Biological Station, University of Helsinki, 16900 Lammi, Finland
169 19. German Environment Agency, Unit Drinking Water Resources and Water Treatment, Corrensplatz 1, 14195 Berlin, Germany
20. Department of Biology, Limnological Institute, University of Konstanz, 78464 Konstanz, Germany
21. Department of Experimental Limnology, Leibniz Institute of Freshwater Ecology and Inland Fisheries, 16775 Stechlin, Germany
22. Institute of Biochemistry and Biology, Potsdam University, 14469 Potsdam, Germany
23. Water Quality Department, Athens Water Supply and Sewerage Company, 11146 Athens, Greece
24. Institute of Nanoscience and Nanotechnology, National Center for Scientific Research «DEMOKRITOS», 15341 Attiki, Greece
25. Department of Botany, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece
26. Centre for Freshwater and Environmental Studies, Dundalk Institute of Technology, A91 K584 Dundalk, Ireland
27. Department of Sustainable Ecosystems and Bioresources, Fondazione Edmund Mach, 38010 San Michele all’Adige, Italy
28. Institute of Botany, Nature Research Centre, Vilnius 08412, Lithuania
29. Department of Freshwater Ecology, Norwegian Institute for Water Research, 0349 Oslo, Norway
30. Department of Hydrobiology, University of Bialystok, 15245 Bialystok, Poland
31. Institute of Environmental Protection and Engineering, University of Bielsko- Biala, 43309 Bielsko-Biala, Poland
32. Institute of Technology, The State University of Applied Sciences, 82300 Elblag, Poland
33. Department of Marine Biotechnology, University of Gdansk, 81378 Gdynia, Poland
170 34. Department of Animal Biology, Plant Biology and Ecology, University of Jaen, 23701 Jaen, Spain
35. Institute of Nature Conservation, Polish Academy of Sciences, 31-120 Krakow, Poland
36. European Regional Centre for Ecohydrology of the Polish Academy of Sciences, 90364 Lodz, Poland
37. Department of Hydrobiology and Protection of Ecosystems, University of Life Sciences in Lublin, 20262 Lublin, Poland
38. Department of Icthyology, Hydrobiology and Aquatic Ecology, Inland Fisheries Institute, 10719 Olsztyn, Poland
39. Department of Water Protection Engineering, University of Warmia and Mazury, 10-720 Olsztyn, Poland
40. Department of Tourism, Recreation and Ecology, University of Warmia and Mazury, 10-720 Olsztyn, Poland
41. Department of Water Protection, Adam Mickiewicz University, 61614 Poznan, Poland
42. Department of Hydrobiology, Adam Mickiewicz University, 61614 Poznan, Poland
43. Institute of Environmental Engineering, Poznan University of Technology, 60965 Poznan, Poland
44. Faculty of Biology, University of Warsaw, 02-096 Warsaw, Poland
45. Department of Environmental Improvement, Faculty of Civil and Environmental Engineering, Warsaw University of Life Sciences—SGGW, 02-787 Warsaw, Poland
46. Department of Hydraulic Engineering, Faculty of Civil and Environmental Engineering, Warsaw University of Life Sciences—SGGW, 02-787 Warsaw, Poland
47. Department of Freshwater Protection, Institute of Environmental Protection- National Research Institute, 01-692 Warsaw, Poland
171 48. Department of Plant Ecology and Environmental Conservation, Faculty of Biology, University of Warsaw, 02-089 Warsaw, Poland
49. Centro de Investigação da Montanha, Instituto Politécnico de Bragança, Campus de Santa Apolónia 5300-253 Bragança, Portugal
50. Instituto Dom Luiz, University of Lisbon, 1749016 Lisbon, Portugal
51. Institute of Earth Sciences Jaume Almera, ICTJA, CSIC, 08028 Barcelona, Spain
52. Interdisciplinary Centre of Marine and Environmental Research (CIIMAR) and University of Porto, 4450-208 Matosinhos, Portugal
53. Research Center in Biodiversity and Genetic Resources (CIBIO-Azores), InBIO Associated Laboratory, Faculty of Sciences and Technology, University of the Azores, 9501-801 Ponta Delgada, Portugal
54. Faculty of Natural Sciences and Mathematics, SS Cyril and Methodius University, 1000 Skopje, Macedonia
55. National Reference Center for Hydrobiology, Public Health Authority of the Slovak Republic, 82645 Bratislava, Slovakia
56. Department of Water Quality, Slovenian Environmental Agency, 1000 Ljubljana, Slovenia
57. Department of Genetic Toxicology and Cancer Biology, National Institute of Biology, 1000 Ljubljana, Slovenia
58. Department of Civil Engineering, University of A Coruña, 15192 A Coruña, Spain
59. Department of Evolutionary Biology, Ecology, and Environmental Sciences, University of Barcelona, 08028 Barcelona, Spain
60. Department of Limnology and Water Quality, AECOM U.R.S, 08036 Barcelona, Spain
61. Department of Biology, University of Cádiz, 11510 Puerto Real, Cádiz, Spain
62. Catalan Institute for Water Research (ICRA), 17003 Girona, Spain
63. Department of Ecology, University of Granada, 18071 Granada, Spain
64. R&D Department Environmental Engineering, 3edata, 27004 Lugo, Spain
172 65. Department of Ecology, University of Malaga, 29071 Malaga, Spain
66. Centro de Investigacións Cientificas Avanzadas (CICA), Facultade de Ciencias, Universidade da Coruña, 15071 A Coruña, Spain
67. Cavanilles Institute of Biodiversity and Evolutionary Biology, University of Valencia, 46980 Paterna Valencia, Spain
68. Department of Microbiology and Ecology, University of Valencia, 46100 Burjassot, Spain
69. Department of Biology, Lund University, 22362 Lund, Sweden
70. Department of Forest Engineering, University of Cankiri Karatekin, 18200 Cankiri, Turkey
71. Department of Ecology and Genetics, Limnology, Uppsala University, 75236 Uppsala, Sweden
72. Department of Ecology and Genetics, Erken Laboratory, Uppsala University, 76173 Norrtalje, Sweden
73. Department of Aquatic Ecology, Netherlands Institute of Ecology (NIOO- KNAW), 6700 Wageningen, The Netherlands
74. Department of Environmental Sciences, Wageningen University & Research, 6700 Wageningen, The Netherlands
75. Department of Environmental Sciences, Aquatic ecology and water quality management group, Wageningen University, 6700 Wageningen, The Netherlands.
76. Society for the Protection of Prespa, 53077 Agios Germanos, Greece
77. Institute for Water and Wetland Research, Department of Aquatic Ecology and Environmental Biology, Radboud University Nijmegen, 6525 AJ Nijmegen, The Netherlands
78. Research Institute RIKILT, BU Contaminants & Toxins, Wageningen University, 6708 WB Wageningen, The Netherlands 79. Laboratoire Microorganismes - Génome et Environnement, Université Blaise Pascal - Clermont-Ferrand II, Aubiere Cedex 63177, France
173 80. Department of Biological Sciences, Virginia Tech, 24061 Virginia, USA 81. Institute of Marine Sciences, University of North Carolina at Chapel Hill, 28557 North Carolina, USA 82. National Institute of Environmental Health, 1097 Budapest, Hungary 83. Department of Molecular Biology and Genetics, Gaziosmanpasa University, 60250 Merkez, Turkey 84. Department of Fisheries and Aquaculture, Ankara University, 6100 Ankara, Turkey 85. Department of biology, Middle East Technical University, 6800 Ankara, Turkey 86. Department of Biology, Balikesir University, 10145 Balikesir, Turkey 87. Institute of Marine Sciences, Department of Oceanography, Middle East Technical University, 06800 Ankara, Turkey 88. Department of Environmental Engineering, Abant Izzet Baysal University, 14280 Bolu, Turkey 89. Department of Environment and Resource Management, Mediterranean Fisheries Research Production and Training Institute, 7090 Antalya, Turkey 90. Department of Bioengineering, Bursa Technical University, 16310 Bursa, Turkey 91. Department of Biology, Hitit University, 19040 Corum, Turkey 92. Department of Basic Science, Ataturk University, 25240 Erzurum, Turkey 93. Department of Biology, Giresun University, 28100 Giresun, Turkey 94. Republic of Turkey Ministry of Food Agriculture, Fisheries Research Institute, 32500, Eğirdir, Isparta, Turkey 95. Department of Freshwater Resource and Management, Faculty of Aquatic Sciences, Istanbul University, 34134 Istanbul, Turkey 96. Faculty of Aquaculture, Mersin University, 33160 Mersin, Turkey 97. Institute of Marine Sciences, Marine Biology and Fisheries, Middle East Technical University, 33340 Mersin, Turkey 98. Department of Biology, Sakarya University, 54187 Sakarya, Turkey 99. Department of Biological and Environmental Sciences, University of Stirling, FK9 4LA Stirling, United Kingdom 100. School of Pharmacy and Life Sciences, Robert Gordon University, AB10 7GJ Aberdeen, United Kingdom
174 101. Research department for Limnology, University of Innsbruck, A-6020 Innsbruck, Austria 102. Independent graphic designer and illustration, 6a Everton Ave, Cabra, D07 PN2W Dublin, Ireland 103. Independent graphic designer, 11 Aravon Ct, A98 D529 Bray, Ireland
175 CV & List of Publications
PhD candidate Evanthia Mantzouki is a qualified researcher in Aquatic sciences with a special desire to unite the scientists of the world towards a common goal: a better freshwater quality for ecological and societal purposes. During her PhD project, she coordinated the European Multi Lake Survey to study the effects of eutrophication and global warming on cyanobacterial and cyanotoxin occurrence, in hundreds of lakes at the continental scale. Ultimately, she wishes she will lead projects where scientists, citizens and stakeholders work together to establish sustainable environmental approaches.
Education
April 2014 - October 2018 PhD in Microbial Ecology
University of Geneva, Switzerland
February 2012 - March 2014 MSc in Limnology and Oceanography
University of Amsterdam, The Netherlands
September 2005 - July 2011 Diploma in Biology
Aristotle University of Thessaloniki, Greece
Publications
Mantzouki, E. and Ibelings, B.W. 2018. The Principle and Value of the European Multi Lake Survey. Limnology and Oceanography bulletin, 27(3):82-86. Mantzouki, E.; ... and Ibelings, B.W. 2018. A European Multi Lake dataset of phytoplankton, cyanotoxins and environmental parameters. Nature Scientific Data. Mantzouki, E.; ... and Ibelings, B.W. 2018. Temperature effects explain continental scale distribution of cyanobacterial toxins. Toxins, 10(4):156. Suarez-Bolanos, E.L.; Tiffay, M.-C.; Chekryzheva, T.; Kalinkina, N.; Sharov, A.; Makarova, E.; Zdorovennon, R.; Mantzouki, E.; Syarki, M.; Tekanona, E.; Venail, P.; and Ibelings, B.W. Diurnal variation in convection-driven vertical distribution of the phytoplankton under ice and after ice-off in the large Lake Onego (Russia). Accepted in Inland Waters.
176 Mantzouki, E.; ... and Ibelings, B.W. 2018. A continental-scale multi-lake survey in Europe’s third hottest summer: cooler Boreal regions develop bigger cyanobacterial blooms than warmer Mediterranean regions. Submitted to Global Climate Change. Mantzouki, E.; Weyhenmeyer, G.; Grossart, H.P.; Brookes, J.; Beklioglu, M.; deSenerpont Domis, L.; Dugan, H.; Skaff, N.; Dubek, J.; Nejstgaard, J.; Seelen, L; Sadro, S.; Teubner, K.; and Ibelings, B.W. 2018. Opinion: Snapshot Surveys for lake monitoring – more than a shot in the dark. Accepted in Frontiers in Ecology & Evolution. Mantzouki, E.; Visser, P.M.; Bormans, M. and Ibelings, B.W. 2016. Understanding the key ecological traits of cyanobacteria as a basis for their management and control in changing lakes. Aquatic Ecology, 50(3):333-350.
177 EMLS Dataset The full dataset can be accessed online: https://doi.org/10.6073/pasta/dabc352040fa58284f78883fa9debe37 (2018)
178