Program

Keynote Speakers

K1

Maarja Kruusmaa
Tallinn University of Technology, Estland

Talk: Aquatic sensing using underwater robots: preparing for a deluge of environmental data.

K2

Marie A. Roch

San Diego State University, USA

Talk: Mining and management of eco-acoustic data.

K3

Debra P. C. Peters
United States Department of Agriculture (USDA) & New Mexico State University, Las Cruces, USA
Talk: Harnessing the power of big data in ecology by machine learning.

K4

Duccio Rocchini

University of Trento, Italy

Talk: A look beyond biogeographical theories: advances and challenges.

This page shows an overview of the conference themes, keynote talks, regular and special sessions. To see more information for every item please click on the ‘+’

Theme 1: Ecological Monitoring and Data Management

Keynote Talks

K1 _ Aquatic sensing using underwater robots: preparing for a deluge of environmental data. (Maarja Kruusmaa, Estonia)

Information is everywhere. But in some places it is incredibly difficult to obtain. This talk tells a story of how to get information from water. The exploration begins by taking ideas from highly advanced biological systems to build human-purposed technologies, and presents new types of bio-inspired sensors that pick up signals from water never before perceived. It then investigates various ways of modelling, interpreting, classifying flow information to better understand what is going on in the aquatic realm, by exploiting the interactions occurring at the interface between a fluid and solid. To conclude, the talk illustrates several real-world research applications where flow information can be used to better understand, protect, control and explore within the underwater world.

Maarja Kruusmaa is one of Europe’s leaders in novel multi-sensor monitoring of ecosystems by novel robots. Her research interests are robotics, biorobotics, artificial muscles, electroactive polymers, underwater robotics, robot learning, flow sensing and experimental fluid dynamics. Maarja received her Ph.D. degree from Chalmers Univeristy of Technology, Gothenburg, Sweden and is now Professor at Tallinn University of Technology, Estland. More information can be found at: https://www.ttu.ee/institutes/centre-for-biorobotics/centre/people-7/

K2 _ Mining and management of eco-acoustic data. (Marie Roch, USA)

Long-term passive acoustic data sets can provide insights into many topic areas related to animals such as behavior, ecology, density, and communication. Acoustic data sets are becoming ubiquitous as the cost of acquisition, storage, and analysis hardware decreases coupled with an explosion of scientific data analysis products. Advances in algorithms to classify and localize sounds are permitting us to ask questions that would not have been possible only a short time ago.
In this talk, we will examine work from projects that discuss classification in the context of large data sets. We will also examine the management of acoustic recording metadata. Many of the processes that affect animal populations operate on multiple time scales (e.g., diel, lunar, decadal) or over large spatial areas. We present methods that can be used to retain and organize these metadata and place them in context with respect to other data such as lunar phase or climatic indicators.

Marie A. Roch is interested in pattern recognition, particularly as applied to categorization of audio signals. Her current work is in pattern recognition for bioacoustics, the study of sound production and perception in animals. She studies questions of identification, behavior, and communication through acoustics. Marie received her Ph.D. degree from The University of Iowa, USA, and is now Professor at San Diego State University, USA. She is associate editor of Ecological Informatics for Imaged Based Monitoring. More information can be found at: http://roch.sdsu.edu/

Regular Sessions

R1.1 _ Ecological monitoring by camera, thermal and acoustic images. (Session Chair: Marie Roch)

Camera traps, thermal infrared videos and soundscapes are non-invasive monitoring techniques of elusive animals without significantly sacrificing analytical accuracy. These methods reduce field hours for estimating demographic parameters, inventory species and migration patterns. Automatic classification systems of camera, thermal and acoustic images allow large datasets to be analyzed over short timescales, and yield valuable information for natural resource decision-making.

This session welcomes papers on all aspects of automatic acquisition, classification and interpretation of image-based data of biodiversity.

R1.2 _ GIS visualisation and communication of remotely sensed ecological data. (Session Chair: Duccio Rocchini)

In spatial ecology synthesizing and properly visualising data in 2D systems is a key issue when aiming at explaining spatial patterns by spatial processes.

This has been demonstrated in a number of ecological and geographical studies, with different scientific aims. Increasing availability of open ecological data through networks like the Global Biodiversity Information Facility or the DataONE (Data Observation Network for Earth), on the one hand, and remote sensing data on the other, makes it necessary to promote methods for data visualization and communication.

This session is devoted to methods for ecological data synthesis, relying on challenging GIS and remote sensing approaches.

Special Sessions

S1.1 _ Earth Observation for ecosystem analysis and decision making. (Session Chairs: Klaus Joehnk and Janet Anstee)

Earth observation data is acquired from satellites, airborne platforms, and in-situ measurements on land and in water. Combined with modelling tools this yields detailed insights into ecosystem functioning from small to large scales and allows for improved prediction algorithms, which then provides decision makers with reliable and up-to-date information.

In this special session we want to address the challenges and opportunities
of the use of Earth observation data in modelling and forecasting of ecosystem trends. This can be, but is not restricted to, water quality modelling, harmful algal blooms, biodiversity trends, ecosystem accounting, etc.

Keywords: earth observation, remote sensing, water quality, bio-optics

S1.2 _ Analysis of ecoacoustic recordings: detection, segmentation and classification. (Session Chairs: Jérôme Sueur and Dan Stowell)

Ecoacoustics is a newly emerged discipline that aims at tackling ecological research questions through the lens of sound analysis. Ecoacoustics covers several questions in marine, freshwater and terrestrial environments dealing with biodiversity monitoring, population ecology, community ecology and landscape ecology. One of the key approaches of ecoacoustics consists in identifying sounds of ecological importance in environmental recordings that were collected in an unattended way by automatic recorders. This search task is made difficult by the occurrence of background noise due to human activities, the co-occurrence of several sounds of interest, the degradation of the sounds of interest related to their propagation in the environment, a high-degree of variability of the sounds of interest, a large amount of data, and a lack of reference archives. Solutions including computer processes are currently in development to try to get around these difficulties. This session will be the occasion to report and share new techniques involving signal analysis, machine learning, deep learning and high dimension statistics for advances in detection, segmentation, supervised and unsupervised classification of sound events.

Keywords: ecoacoustics, sound analysis, classification

S1.3 _ Computer vision in environmental sciences. (Session Chairs: Shinji Fukuda and Jeffrey Tuhtan)

Image-based methods are at the forefront of artificial intelligence applications. This special session provides a forum for researchers and professionals using image-based methods to study species, population, biodiversity and the abiotic environment.

The topics of this special session include:
-UAV imagery
-GIS/orthoimagery
-Video tracking/motion estimation
-Object recognition and classification
-High speed imaging
-Multispectral remote sensing

Keywords: image analysis, machine learning, artificial intelligence, classification/regression

S1.4 _ Integrating data for analysis – how far are we? (Session Chairs: Corinna Gries, Margaret O'Brien, Kristin Vanderbilt and Colin Smith)

This session will discuss advances and tools for the second step of the data life cycle in ecological research. The first step being either field data collection and/or data discovery, the second step has been called the ‘janitorial’ step, but is mostly known as data cleaning, data harmonization, or data integration. It poses major data management challenges and is frequently a time consuming process with estimates of up to 80% of the data analysis. The reasons for this large proportion of effort range from those that cannot be addressed with technological solutions and are rooted in sampling methods to those that are related to data organization and semantics that may be addressed with developing technologies.
This session will explore the progress that is being made toward reducing the effort needed for pre-analysis data harmonization. Encouraged are: (1) reports on data integration projects spanning the range of employing and advancing semantics, ontology, linked data, specific tools, workflow systems, and standards developments, (2) considerations of an approach’s promise for a high return on the investment and/or whether it will it significantly improve documentation of data manipulations, (3) experiences and discussions focusing on comparing effectiveness in reducing time spent in data integration, (4) technological gaps and shortcomings.

Keywords: data synthesis, data integration, data harmonization

S1.5 _ The German Federation for Biological Data (GFBio) - from data acquisition to analysis. (Session Chairs: Michael Diepenbroek, Birgitta König-Ries and Anton Güntsch)

The German Federation for Biological Data (GFBio) aims to set up a sustainable, service oriented, national data infrastructure facilitating data sharing and stimulating data intensive science in the fields of biological and environmental research. GFBio follows a holistic approach including technical, organizational, cultural, and policy aspects. The development of the infrastructure is essentially based on the collective experience and expertise of leading researchers from multiple disciplines as well as on a network of complementary and professional data facilities in the biological and environmental sciences communities, including PANGAEA, major German natural history collection data repositories, and selected facilities from the molecular biology research community. GFBio is projected for three
phases setting out the way from development to management of services.

Keywords: data aquisition, data archiving, data discovery, terminologies, data integration, data visualization, data analysis

S1.6 _ Semantics for biodiversity and ecosystem research. (Session Chairs: Alsayed Algergawy, Clement Jonquet, Naouel Karam, Friederike Klan, Nikos Minadakis, Alessandro Oggioni and Ilaria Rosati)

Biodiversity research aims at comprehending the totality and variability of organisms, their morphology, genetics, life history, habitats and geographical ranges; including the network of interactions with the abiotic and biotic components. Ecosystem research puts its focus on how natural systems and their valuable resources can be protected and thus is tightly coupled to biodiversity. Both domains are outstanding not only with respect to their societal relevance, but also from a data science point of view. They deal with heterogeneous and distributed data resources generated from a large number of disciplines which need to be integrated to advance scientific knowledge in these areas. The presence of such a myriad of data resources makes integrative biodiversity and ecosystem research increasingly important, but at the same time very challenging. It is severely strangled by the way data and information are made available and handled today. Semantic Web techniques have shown their potential to enhance data interoperability, discovery and integration by providing common formats to achieve a formalized conceptual environment, but have not been widely applied to address open data management issues in the biodiversity domain as well as in ecosystem research.
This session aims at bringing together computer scientists, biologists and ecologists working on Semantic Web approaches in biodiversity and ecosystem research, including related areas such as agro-ecology. After the successful of a number of initiatives of the organizers, such as the “Thesauri & Semantics in the Ecological Domain”, “Ontology & Semantic Web for Web for Research” and “Semantics for Biodiversity” workshops, the goal of the session is to keep up the positive momentum and attempt to define a common strategy for advancing semantic web approaches in these domains. The goal is to present new ideas and early on experiences related to the design of high quality biodiversity and ecosystem information systems based on Semantic Web techniques and to foster the exchange on these topics between disciplines.
We welcome topics related to the development and application of semantic technologies to support research in the biodiversity and ecosystem domain and related areas. These include, but are not limited to the following areas:
· Applications of Semantic Web technologies for biodiversity
· Semantic data integration
· Development and design of domain specific ontologies
· Ontology-based applications
· Semantic annotation of biodiversity data
· Semantic approaches for the discovery of biodiversity data and research data services
· Semantic support for scientific workflows
· Data provenance and reproducibility
· Data lifecycle management
· Knowledge extraction and text mining
· Ontology learning
· Standards for biodiversity Data
· Linked Open biodiversity Data
· Ontology development for biodiversity
· Semantic representation of biodiversity and ecosystem data
· Interoperability of biodiversity and earth observation data

Keywords: semantic web, biodiversity data, ecosystem data, integrative research, semantic annotation, semantic data integration, semantic data interoperability, ontology based applications

Theme 2: Ecological Data Synthesis by Machine Learning and Hybrid Modelling

Keynote Talk

K3 _ Harnessing the power of big data in ecology by machine learning. (Debra P. C. Peters, USA)

My talk will be about a Trans-Disciplinary Data-Model Integration (TDMI) approach that focuses on spatio-temporal modeling and cross-scale interactions, and employs interactive machine learning strategies. Applied to ecological problems, my approach integrates knowledge and data on: (1) biological processes, (2) spatial heterogeneity in the land surface template, and (3) variability in environmental drivers using data and knowledge drawn from multiple lines of evidence (i.e., observations, experimental manipulations, analytical and numerical models, products from imagery, conceptual model reasoning, theory). I will apply this approach to a suite of increasingly complex ecologically-relevant problems, and will show how the framework can be linked with a Data Science Integration System (DSIS) to allow more complex questions to be addressed in the future.

Debra P. C. Peters does research in Systems Biology, Ecology and Bioinformatics. Her expertise is amongst others in Ecosystem Functioning, Conservation Biology, Natural Resource Management, Climate Change as well as GRASS GIS, Big Data, Simulation Modeling and Machine Learning. Debra received her Ph.D. degree from Colorado State University, USA. She currently works as Senior Advisor for Earth Observations at the Agricultural Research Service (ARS), United States Department of Agriculture (USDA), and is a Research Associate of the New Mexico State University, Las Cruces, USA. More information can be found at: https://www.researchgate.net/profile/Debra_Peters

Regular Sessions

R2.1 _ Understanding species distribution, population dynamics and phenology by machine learning. (Session Chairs: Antonino Staiano, Alberto Lanzoni and Friedrich Recknagel)

Inferential modelling by machine learning techniques allows efficient and user-friendly analysis and synthesis of highly complex ecological data. Methods like random forest, quantile regression forest, Maxent, GARP have been successfully applied for species distribution modelling of extensive spatial data resulting in species response curves that describe a species’ response to a given habitat condition, and illustrate specific habitat requirements for the species. Non-supervised artificial neural networks, regression trees and canonical correspondence analyses allow to ordinate and classify complex ecological data. Applications of support vector machines, supervised artificial neural networks and evolutionary algorithms to large spatial and temporal data allow to predict population dynamics and reveal phenology.

This session welcomes papers on all aspects of inferential modelling of ecological data by means of novel machine learning techniques.

R2.2 _ Integrated modelling of ecosystem services for decision support in environmental management and policy development. (Session Chair: Peter Goethals)

Ecosystem services modeling integrates aspects of ecosystem use and management, to diverse conversion processes in ecosystems, the resulting goods and services, as well as the social-economic consequences. A challenge is the integration of such diverse processes and components. This session is open to solutions to such integrated modeling challenges, ranging from integrated mechanistic models, to hybrid models and black box approaches.

Special Sessions

S2.1 _ Computational Bioindication for threshold identification in natural resource management. (Session Chairs: Tao Tang, Qinghua Cai and Peter Goethals)

Many novel threshold methods have been brought forward in recent years, partly due to advances in computation and modeling techniques. These methods are extensively used to identify thresholds for univariate biological metrics or multivariate community composition data for terrestrial, freshwater, and marine ecosystems. Thresholds have been applied to determine criteria for biodiversity conservation, biological reference conditions, ecosystem management, natural resource damage assessments, etc. Nevertheless, ecological threshold development still needs more theoretical and empirical research for better understanding. The most urgent issue is that prerequisites for threshold identification should be clearly specified to prevent possible misuse. Many statistical models can detect and quantify thresholds no matter whether a “true” or ecologically meaningful threshold exists or not. It is important to conduct data exploration to obtain preliminary information on relationships between (among) the study variables before identifying thresholds. For instance, we need to know whether causal relationships are relevant or not; whether nonlinear models are better than linear ones for fitting the relationships; whether confounding effects from other environmental factors should be accounted for when quantifying threshold to the target factor(s), etc. Moreover, it is also important to assess whether the identified ecological thresholds are attainable targets through effective management practice. This special session will review the state-of-art of knowledge on ecological thresholds to inform the audience what we know and what key challenges are still open. More importantly, the speakers will share their ideas on the abovementioned topics for better understanding of threshold theories and methods. This session will facilitates communication and collaboration among researchers by providing a forum to think about ecological thresholds from a scientifically sound perspective.

Keywords: threshold response, threshold methods, causal relationship, environmental standards

S2.2 _ Why Bayesian? Integrating environmental modelling with Bayesian inference techniques. (Session Chairs: George Arhonditsis, Alex Neumann and Yuko Shimoda)

The credibility of the scientific methodology of environmental models and their adequacy to form the basis of public policy decisions have been frequently challenged. The current challenges make compelling the development of more realistic modeling platforms (i) to elucidate causal mechanisms, complex interrelationships, direct and indirect ecological paths; (ii) to examine the interactions among the various stressors (e.g., climate change, urbanization/land‐use changes, alternative management practices, invasion of exotic organisms); and (iii) to assess their potential consequences on ecosystem functioning. The proposed session aims to provide insights into the current state of the field, and also highlight the major challenges and future directions of research. Special emphasis will be placed on studies that address topics, such as novel uncertainty analysis techniques, Bayesian inference methods (including Bayesian networks), development of new model formulations and proper representation of biotic functional types, emerging techniques of data assimilation and model optimization, effective integration of physics with biology, and strategies to improve the contribution of complex models to ecological theories. The proposed session encourages contributions from both mathematical and statistical ecosystem modelers.

Keywords: bayesian inference, uncertainty analysis, environmental management, policy analysis

S2.3 _ Quantifying the functions in terrestrial ecosystems: from concepts to data driven methods. (Session Chairs: Talie Musavi, Miguel Mahecha, Xuanlong Ma and Mirco Migliavacca)

One argument for nature conservation efforts is preserving e.g. ecosystem services such as carbon sequestration potential, water and nutrient retention, among many others. To quantify such processes, we not only need a clear conceptual understanding of ecosystem functioning
but also novel ways to quantify them and understand controlling factors at various spatial and temporal scales.

In this session we invite contributions on conceptual advances to define and identify “Ecosystem Functioning”. We also aim to discuss latest developments in observing ecosystem functions from in-situ data or with proximal or remote sensing from the site level to the global scale. Innovative advances on nonlinear statistical methods, model-data integration, or inversion studies that help us to constrain ecosystem functioning or to retrieve functional properties of the ecosystems (e.g.
radiation use efficiency, nitrogen use efficiency, water use efficiency and so forth) – even if in early stages of development are also encouraged to participate as we strive for a broad session. The goal of our session is to provide a stage for those interested in understanding the ecosystem functioning combining diverse data streams or analytic frameworks.

Keywords: ecosystem function, ecosystem processes, model, in-situ data, remote sensing

S2.4 _ Deep learning for environmental science and ecology. (Session Chairs: Christian Requena-Mesa, Basil Kraft, Markus Reichstein, Joachim Denzler and Marco Körner)

Deep learning is an extremely active research area in machine learning and pattern recognition communities. It has gained huge success in areas such as speech recognition, computer vision, or natural language processing. Applications in geosciences and ecology, like extracting knowledge from big-data, short-term forecasting or anomaly detection are promising, in particular since deep-learning can deal very well with space-time structures.

In this session, we invite contributions on the use of deep learning in ecology and environmental science in a series of talks lasting up to half a day (3-4 hours –1 keynote and 11-15 talks). We aim to discuss the latest developments in deep learning for insight into and prediction of ecological systems. We welcome contributions covering all aspects of ecology and environmental science, including biodiversity, climate impact, ecosystem and organismal ecology, biogeography etc.

Keywords: deep learning, machine learning, big-data, environmental science, biodiversity, organismal ecology, community ecology, ecosystem ecology

Theme 3: Applications, Innovations and Communication of Eco-informatics

Keynote Talk

K4 _ A look beyond biogeographical theories: advances and challenges. (Duccio Rocchini, Italy)

Biogeographical theories on species distributions and the spatial variation of biodiversity in space and time represent a long lasting theme in ecology. The development of new modelling techniques based on spatial science and remote sensing allow nowadays to test them in a theoretical (virtual) and an empirical domain.

In this talk I will retrace the progress in ecological informatics applied to biogeography, focusing on exciting advances and major challenges.

Duccio Rocchini is Professor in Biology and Ecology at the University of Trento, Italy. He received his Ph.D. degree from the University of Siena, Italy. His main research interests are related to species distribution modelling, spatial and computational ecology and ecological remote sensing. Over the years, he promoted the use of remote sensing for the study of biodiversity change in space and time, publishing more than 120 ISI papers on this theme. He is currently Associate Editor of Ecological Informatics, being responsible for the “GIS, Remote Sensing and Biogeography” theme. More information can be found at: https://www4.unitn.it/People/en/Web/Persona/PER0196062#INFO

Special Sessions

S3.1 _ THIS SESSION HAS BEEN CANCELLED _ Eco-metabolomics meets eco-informatics: Identifying ecologically relevant compounds in metabolomics datasets. (Session Chairs: Nicole van Dam, Christoph Steinbeck and Steffen Neumann)

– This session has been cancelled –

Metabolomics is more and more commonly applied to a wide range of non-model organisms that are studied by ecologists. This means that there is an increasing number of ‘unknown’ compounds for which there are -high quality- mass spectral data. Due to the mere numbers of features that are detected by metabolomics, it is barely possible to identify each unknown manually. There are several, independent bioinformatics groups, for example in Germany at FSU Jena (Sirius, CSI:FingerID) and IPB Halle (MetFrag, MetFamily), in Spain (iMet), Israel (WEIZMASS) and elsewhere, developing databases, software packages and workflows which help to (partially) identify or at least annotate compounds. At the same time, ecologists are increasingly interested to apply these workflows to identify biological active metabolites for example in plant materials and root exudates. In this session we would like to bring the bioinformaticians developing the software and databases together with the ecologists interested in applying the developed programs to their metabolomics datasets. The aim would be to identify the potential and limitations of the currently developed tools, to motivate ecologists to apply the packages as well as to upload their metabolomics datasets to the relevant databases.

Keywords: metabolomics, bioinformatics, mass spectrometry, software programs, chemical ecology, phytochemistry

S3.2 _ Citizen science meets informatics: Data science challenges in ecological research with public participation. (Session Chairs: Friederike Klan and Jana Wäldchen)

During the last decade, Citizen Science, i.e. the involvement of laymen in scientific research, has gained great attention, both from the public and within the scientific community. Particularly the life sciences benefit from this development as citizen scientists contribute environmental observations of high resolution, analyze large amounts of ecological data or raise entirely new research questions. In doing so, they help to tackle pressing societal challenges such as loss of biodiversity and climate change.

In addition to the issue of how to engage and empower volunteers, data science aspects are major challenges in Citizen Science projects. This includes the following questions
1.How to make sure that data collected by citizen scientists are useful and relevant for addressing scientific questions?
2.How can data collected by citizen scientists be found, accessed, interpreted and used by others?
3.How to integrate and analyze data collected by the public with other data sources?
4.How to assess and improve the quality and reliability of Citizen Science data?
5.How to increase the credibility of data collected by volunteers and how to acknowledge citizen contributions?
6.How to enable citizen scientists to gain insights from data?

While some of these topics are specific to Citizen Science, they often share challenging aspects of data-intensive science in general (e.g. How to make data findable, accessible, interoperable and re-usable?). However, despite first cross-disciplinary and Citizen Science specific initiatives such as FORCE11 and the FAIR Data Principles [1], the Cost Action “Citizen Science to promote creativity, scientific literacy, and innovation throughout Europe” [2] or first ideas on a EU Citizen Science Gateway for Biodiversity Data [3] where basic data science challenges are jointly discussed and best practices are collected, Citizen Science practitioners often address these core questions in an ad-hoc manner and individually in the context of specific projects.

We believe that the Citizen Science community would greatly benefit from an intensified scholarly exchange on data science topics related to Citizen Science across projects and disciplines. Thus, the objective of this special session is to bring together practitioners in Citizen Science projects as well as “traditional” scientists to discuss basic data science challenges arising in Citizen Science projects, to share best practices and lessons learned as well as to identify next steps towards a regular exchange on these topics and to initiate joint efforts on systematically addressing these challenges.

Keywords: citizen science, data science, data management

S3.3 _ Operationalizing essential biodiversity variables: aspects of data integration, production and dissemination. (Session Chairs: Néstor Fernández, W. Daniel Kissling, Robert Guralnick and Alex Hardisty)

The Group on Earth Observations – Biodiversity Observation Network (GEO BON) has the impetus to develop conceptual and technical approaches to the production of Essential Biodiversity Variables (EBVs). However, the ecological scientific community as well as those who are responsible for acquiring, curating, publishing, processing and using heterogeneous biodiversity and ecological data must invest into supporting a systematic production and use of EBVs. Such information products should be applicable to any geographic area, covering time-period(s) of interest for detecting biodiversity change at policy-relevant time scales, with data that is held in any or multiple repositories, and produced by appropriately skilled persons anywhere in the world. Within constraints of specific data types, EBV information products should be harmonised and comparable at various scales from local to global and across time, such that they can be used to monitor and measure biodiversity change.

By showcasing what has been done so far, and guided by principles of global coordination of biodiversity monitoring, this session intends to foster scientific and technical exchange and build communities of practice to support production, delivery, use and sustainability of Essential Biodiversity Variables (EBVs) data products. This session specifically aims to:
• Showcase the role and need for informatics to develop and support of the production and delivery of Essential Biodiversity Variables (EBVs) information products at scales from local to global;
• Showcase approaches for utilising and sustaining EBV information products and their adherence to FAIR (Findable, Accessible, Interoperable, Re-usable) principles.
• Provide recommendations for the production and dissemination of biodiversity observations under the EBV framework

Keywords: biodiversity monitoring, data interoperability, global change, global observation systems, GEO BON, research infrastructures

S3.4 _ Plant trait based approaches for agricultural landscapes restoration and conservation in drylands: data to knowledge. (Session Chairs: Elizabeth Arnaud, Thomas Evert, Roeland Kindt, Marie Angélqiue Laporte, Sarah Jones, Natalia Estrada Carmona, Tobias Fremout and Riza Batista-Navarro)

Dryland agricultural landscapes are characterised by high levels of poverty, food insecurity, and land degradation. Interventions that simultaneously deliver nutrition benefits and halt or restore degraded ecosystems are critical in drylands for achievement of local and global sustainable development. Plant species and community traits regulate multiple ecological processes linked to nutrition and restoration, which cascade into the provision of ecosystem services important for livelihoods and food security. However, there is no readily accessible evidence base documenting the linkages between plant biodiversity in agricultural landscapes, ecological processes and ecosystem services to guide the selection of plant traits in landscape restoration and conservation planning. Such a service would support ecologically-minded farmers to select on-farm plant combinations that provide multiple ecosystem services and help policymakers provide incentives that stimulate uptake of promising combinations at a farm and landscape level. Existing databases record plant traits, uses, and composition but no database provide agriculturally relevant trait-ecosystem service linkages that could enrich the content of decision-support tools for ecosystem restoration and conservation .

This discussion session will assemble selected presentations around results of current efforts made by the scientific community to document the importance of plant diversity in agricultural landscapes through e.g. field surveys in drylands areas, literature and data mining, and make quality data on priority traits, ecosystem services, conservation goals, and linkages available with the use of common terminology, ontologies and metrics. Time will be saved to discuss some key questions in success, opportunities and challenges for collecting, aggregating, accessing and sharing the required critical mass of data needed for evidence-based decision making will be discussed. A short report will be produced and an interest group created.

Keywords: agricultural landscapes, ecosystem services, landscape restoration, plant trait, plant species communities, ontologies, data mining, drylands

S3.5 _ Plant traits and biogeochemical cycles. (Session Chairs: Jens Kattge, Susanne Tautenhahn, Michael Bahn, Han Wang and Oskar Franklin)

Plant traits extend the range of earth observations to the level of individual organisms, providing a link to ecosystem function and modeling in the context of rapid global changes. However, overcoming the differences in temporal and spatial scales between plant trait data and biogeochemical cycles remains a challenge.

This session will address the role of plant species, biodiversity and adaptation / acclimation / optimality and their connection to the biogeochemical cycles of water, carbon, nitrogen and phosphorus.

We welcome conceptual, observational, experimental and modeling approaches, and studies from the local to the global scale, including e.g. remote sensing observations and novel concepts and tools for the acquisition, management, analysis and synthesis of trait data.

Keywords: plant traits, biogeochemical cycles, functional biogeography, ecosystem modelling, plant adaptation / acclimation / optimality

S3.6 _ Efficient data and workflow management for reaching Sustainable Development Goals (SDGs) targets associated to biodiversity and ecosystems. (Session Chairs: Jan Bumberger, Jitendra Gaikwad and Aidin Niamir)

The Sustainable Development Goals (SDGs) are a collection of 17 global goals, including zero hunger, good health and well-being, climate action, clean water and sanitation, affordable and clean energy, that has been agreed by international communities to be hopefully achieved by 2030. Monitoring progress towards these goals require a reliable data and information which are accessible and reproducible over time and space. Earth Observation (EO) products can potentially address such a need for trusted sources of data to monitor the trends of environmental conditions (i.e. essential variables), and also inform models to predict progress (i.e. indicators) towards policy targets over variety of scenarios. However, efficient management of big earth observation datasets and reproducible modeling workflows remains a challenge.

In this session we want to bring together experts representing broad range of experience in applications related to SDGs with a special focus on biodiversity and ecosystem services. We invite contributions presenting challenges, solutions, cases studies, and best practices dealing with big data and modeling workflows management. We intend to discuss on the list (not excluding) of following topics:
– Efficient data management approaches along the chain of information from field data to derived indicators taking into account the uncertainties
– Standardized and operationalized data quality assurance and fusion approaches for biotic, abiotic and other EO data
– From data to variable, and from variable to indicator workflows for biodiversity and ecosystem services
– Best practices to improve interoperability
– Uncertainty/data reliability

Keywords: Sustainable Development Goals (SDGs), Earth Observation (EO), data management, data quality assurance, data fusion, wodeling workflows management, indicator workflows, biodiversity
ecosystem services, uncertainty/data reliability

S3.7 _ Dimensionality reduction in ecology. (Session Chairs: Miguel Mahecha, Guido Kraemer, Markus Reichstein and Gustau Camps-Valls)

“Dimensionality Reduction” (traditionally known under the term “ordination” in Ecology) is one of the most classic and widely used tools for interpreting and visualizing high-dimensional data in ecology. The relevance of these methods for the field is reflected in the fact that ecologist have often proposed methodological innovations in this area. For instance, in 1978 Williamson proposed the “step-across” method to consider nonlinearities in the data space. This idea was largely overlooked outside ecology and developed independently years later in computer science as “Isometric Feature Mapping” (Isomap, Tenenbaum et al. 2000). For many years, this tool was then quite influential tool for manifold learning. However, in recent years the field of dimensional reduction has continued to develop very rapidly and today a wide range of powerful nonlinear methods for nonlinear dimensional reduction are available, or old ideas have gained momentum with the re-popularization of neural networks in the context of deep learning (see e.g. latest progress on Variational Autoencoders).

In this session we invite contributions to present and critically discuss the latest developments in the field of dimensionality reduction. We want to understand the potential of these new methods for applications in ecology. We invite both computer scientists to present new methods that can be applied in ecology and ecologists to outline their problems that could benefit from novel methods. Contributions from the interface between “ecology” and “computer sciences” are very welcome.

Keywords: dimensionality reduction, ordination, nonlinear high-dimensional data analysis

S3.8 _ Using BEXIS 2 to facilitate science from data collection to data publication. (Session Chairs: Roman Gerlach, Javad Chamanara, Sven Thiel, David Schöne and Nafiseh Navabpour)

This session is dedicated to researchers using the BEXIS 2 research data management platform. We invite short presentations from the user community that showcase the usage of BEXIS 2. Showcases can be related to any data management aspect (e.g. metadata creation, data structure definition, data publication) supported by BEXIS 2. In addition, we encourage users to contribute potential features and improvements that they would like to see in BEXIS 2.

Contributed presentations should, first, describe a specific data management problem, and then demonstrate a solution within BEXIS 2. Besides such showcase we encourage any other experience report. Although presentations are expected primarily from the users community (i.e. researchers), we also expect a number of developers and data managers of BEXIS 2 instances to be present in the session, which may lead to some fruitful discussion. After each talk there will be time for discussion.

This session is intended to provide a forum for the BEXIS community, which in previous years met at the annual BEXIS User and Developer Conference. This year the conference has been suspended in favour of the ICEI conference. However, the session is open to anyone else interested in BEXIS 2.

Keywords: research data management, BEXIS 2 platform, data sharing, data publication, open source software