<p>Good scientific practice requires good documentation and traceability of every research step in order to ensure reproducibility and repeatability of our research. However, with increasing data availability and ability to record big data, experiments and data analysis become more complex. This complexity often requires many pre- and post-processing steps that all need to be documented for reproducibility of final results. This poses very different challenges for numerical experiments, laboratory work and field-data analysis. The platform Renku (https://renkulab.io/), developed by the Swiss Data Science Center, aims at facilitating reproducibility and repeatability of all these scientific workflows. Renku stores all data, code and scripts in an online repository, and records in their history how these files are generated, interlinked and modified. The linkages between files (inputs, code and outputs) lead to the so-called <span>knowledge graph, used to record the provenance of results and connecting those with all other relevant entities in the project.</span></p><p>We will discuss here several use examples, including mathematical analysis, laboratory experiments, data analysis and numerical experiments, all related to scientific projects presented separately. Reproducibility of mathematical analysis is facilitated by clear variable definitions and a computer algebra package that enables reproducible symbolic derivations. We will present the use of the Python package ESSM (https://essm.readthedocs.io) for this purpose, and how it can be integrated into a Renku workflow. Reproducibility of laboratory results is facilitated by tracking of experimental conditions for each data record and instrument re-calibration activities, mainly through Jupyter notebooks. Data analysis based on different data sources requires the preservation of links to external datasets and snapshots of the dataset versions imported into the project, that is facilitated by Renku. Renku also takes care of clear links between input, code and output of large numerical experiments, our last use example, and enables systematic updating if any of the input or code files are changed.</p><p>These different examples demonstrate how Renku can assist in documenting the scientific process from input to output and the final paper. All code and data are directly available online, and the recording of the workflows ensures reproducibility and repeatability.</p>

<p>Trees invest carbon in stem growth every year, and allocate the carbon to build xylem vessels, varying in length and diameter, which serve as a path for the transport of water and nutrients from the soil to the leaves.</p> <p>To assess the cost and benefits of carbon investment into xylem and the resulting hydraulic conductivity, we used a combination of two methods. The first was a novel method of measuring hydraulic conductivity under suction using a syringe pump. The second used dye and cryo-microscopy to determine the ratio and dimensions of conducting vessels in individual year rings. Using these methods, we were able to determine that for <em>Fagus sylvatica</em>, larger vessels did not have lower carbon costs per conductivity due to the shorter functional lifespan, whereas the hydraulic conductivity per cross-sectional area was not larger than smaller vessels. In fact, we found a greater wood density for samples with larger median vessel diameter, implying that larger vessels need more carbon for structural support.</p> <p>Here we present these findings and discuss the potential application of the methods to understand how plants adapt their xylem carbon allocation across species and environmental conditions.</p> <p> </p>

<p>Vegetation responds to environmental change in many ways and at various time scales. For example, increasing atmospheric CO<sub>2</sub> concentrations can reduce stomatal conductance and, hence, transpiration at an hourly scale, whereas adjustments in leaf area, photosynthetic capacity and root distributions follow at the daily to seasonal scale. Evidence for root growth plasticity and adaptation to soil moisture conditions can be found in field and experimental data. However, the time scales at which roots respond to a sudden change in soil moisture are not well documented, and the dynamics of root allocation in response to soil moisture changes at daily time scales is not well understood. In addition, when looking at even longer time scales, shifts in tree density and species composition may happen over decades or centuries only. These responses give rise to feedbacks with soil water resources and atmospheric conditions, affecting the entire soil-vegetation-atmosphere system on a large range of spatio-temporal scales.</p><p>Reliable projections of long-term ecosystem response to environmental change require adequate understanding and quantitative representation of the physical processes and biological trade-offs related to vegetation-environment interactions. This includes answering the following questions:</p><p>1) What is the trade-off between canopy CO<sub>2 </sub>uptake and water loss under given atmospheric conditions?</p><p>2) How much carbon do the plants need to invest into their root system, as well as water transport and storage tissues in order to achieve a certain water and nutrient supply for the canopy?</p><p>3) How quickly can root systems respond to changing conditions?</p><p>4) What are the trade-offs between carbon investments into foliage, stems and roots and returns in terms of carbon uptake by photosynthesis?</p><p>5) Do plants adapt to the environment in an optimal way in order to maximise their net carbon profit, i.e. the carbon uptake minus carbon invested into tissues needed for its uptake?</p><p>6) And finally, can vegetation behaviour be predicted by assuming a community-scale optimal adaptation for maximum net carbon profit?</p><p>Here we present promising results related to Question 6) based on the Vegetation Optimality Model (VOM), which was recently applied and tested along a precipitation gradient in Australia. We also explain the benefits of quantitative answers to Questions 1-4 and point to targeted experiments needed to address these questions, some of which will be presented separately.</p>

<p>The goals of open science include easy reproducibility of research results, transparency of research methods and re-usability of artefacts, e.g. data, code, and graphics. Consequently, open science is expected to foster scientific collaboration and sustainability of research, as it enables building on each others' methods and results for many years and decades to come.</p><p>Here we report about our collective attempts in the last 4-10 years of taking open science to the extreme by using exclusively open formats, open-source software, sharing all stages of our work online and recording workflows and provenance of code and data. Most of our analyses are carried out in Jupyter Notebooks, which are all shared online through gitlab. In these notebooks and our python-analyses, we integrate the python package essm for transparent and easily reproducible mathematical derivations. For more complex analyses, including large model runs, we use the tool Renku of the Swiss Data Science Center in order to record workflows and provenance of code and data.</p><p>Find out where we succeeded, where we failed, what we gained and what we lost in pursuing open science to the extreme. Hear about the views and experiences with open science at the undergraduate, postgraduate, postdoc, engineer and senior researcher level. Eventually, we will also report about what we are still missing for entirely reproducible, verifiable, and reusable open science. We hope we can foster a debate about good open science practices, and how we can remove obstacles that are still in our way.</p>

<p>Commonly, xylem hydraulic conductance is measured by applying a positive pressure (above atmospheric) to push water through a twig. To imitate flow in twig samples under natural conditions, we developed a method that applies a controlled flow rate using suction, similar to transpiration-driven flow in plants.</p><p>The setup consists of a syringe pump to control water flow, where a twig is inserted in the flow path and hydraulic conductivity is calculated from measurements using pressure sensors and a flow meter. The syringe pump can be used to generate controlled flow rates in both directions and a series of bypasses can be used to self-calibrate the sensors and reverse flow directions through the twig while the syringe pump is either pushing or pulling. In this way, we were able to compare our suction method with the more conventional pushing method and assess the effect of flow direction on hydraulic conductance measurements. We found a reproducible pattern in measured conductivity values, where measurements using suction resulted in a 50% lower conductivity than when flow was induced by pushing. The direction of flow (root-shoot vs. shoot-root) also had a strong influence, with suction in root-shoot direction resulting in the lowest conductivity measurements, but repeated reversals of flow revealed an intricate pattern of loss and partial restoration of conductivity, implicating the existence of particles that move with the flow and accumulate at the vessel ends.</p><p>Here we present the intriguing results and propose an explanation capable of explaining the reproducible patterns in observed conductivity dynamics during the experiments. The explanation involves nanobubbles that shrink and swell depending on the liquid pressure and surface tension, move with the flow and reduce conductivity as they accumulate at vessel ends.</p>