TOP STORY

Designing Better Experiments in Global Change Ecology

Rapid and unprecedented anthropogenic change is altering multiple environmental factors in the ocean simultaneously. Surface temperatures are rising, ocean pH is reducing, more low-nutrient zones are being created and the ocean is becoming more stratified. These environmental factors have interactive impacts on marine life. Predicting the result of these interactions effectively is crucial to develop informed climate policy.  

Mechanistic mathematical models are critical to predict the impact of anthropogenic change on global oceans.  These models can incorporate multiple environmental drivers and generate predictions about their joint impact. However, the models need the shape of functional form of the interaction between multiple drivers and parameters of the relevant species to generate precise predictions. This information can only be obtained from experiments. Unfortunately, global change experiments usually focus on only one environmental factor at a time. There is good reason for this: experiments with multiple drivers are difficult and resource-intensive. Therefore, we simply don’t know enough about the functional forms of the interactive impacts of multiple drivers of change and the parameters for the relevant species.  

So how do we design feasible yet reliable experiments to predict interactive impacts of multiple factors? One solution is to develop highly efficient experimental designs which generate high amounts of information per measurement. We compared nine experimental designs on their predictive power, a measure of information important to global change ecology.  Out of the nine designs we evaluated, three are commonly used in ecology: present vs future, ANOVA and full factorial response surface designs. A present vs future design compares a specific future scenario (e.g., 28° C and 3 µMol L^-1 nitrogen) with the current scenario (e.g., 22° C and 6 µMol L^-1 nitrogen). An ANOVA design will take this one step further and measure the growth across a 2x2 grid of temperature and nutrient conditions. Finally, a full factorial design measures the growth across a larger grid, typically 4x4 or 5x5. Full factorial designs are resource-intensive but are the gold standard in the field.

An efficient experiment squeezes the most information out of every measurement. Researchers in statistics have developed methods to calculate optimal designs with a specific goal such as better parameter estimates, predictive ability etc. in mind. While optimal designs are very efficient, they also often require prior information such as a function to be fitted and the parameters in case of non-linear functions. This creates a Catch-22: the design cannot be calculated without the parameters and the parameters cannot be found without doing the experiment.  There are two potential solutions: obtaining parameter values from the literature or using sequential designs. In a sequential design, a small experiment is used to estimate the parameters. These parameters are then used to design an optimal design, resulting in a better estimate of parameters. This process is repeated until the experimenter is satisfied with the resulting fitted curve.

We found that the designs best at predictions were the full factorial, optimal design and sequential design while the present vs future and ANOVA designs resulted in the worst predictive capability. Imagine a scenario where the experiment is measuring the dependence of phytoplankton growth on both temperatures and nitrogen concentrations. Given the same total number of measurements, the resulting data from the optimal and factorial designs is far more useful for prediction than the data from present vs future or ANOVA designs.

In our opinion, optimal and full factorial designs are the most efficient for global change experiments. Unfortunately, optimal designs are virtually unknown in ecology perhaps due to the significant mathematical barriers and need for prior information about the focal species. Therefore, we recommend that experimenters use full factorial designs if they are working with a species for which parameters are unknown. However, if relevant parameter estimates are available, optimal designs can be used to save time and effort. Efficiently designed experiments in combination with reliable mathematical models are the most fruitful way forward to predict global change impacts on the oceans.

IN THE FIELD WITH...

Ilse van Opzeeland: Reviving PALAOA or the Acoustic Homecoming

Slowly the impressive lance sinks into the ice. It reminds of a Jules Verne-like scientific instrument in the way its copper plates are bolted so craftfully. It is attached to an enormous installation with howling diesel-powered engines that breathes a slow primitive power. Maybe exactly the type of primitive power and patience that is needed to melt through 100 metres thick shelf ice. We are in the Antarctic standing on the Ekström Ice Shelf and we are melting holes to finally revive our passive acoustic observatory after a 2-year data gap due to an ice shelf break off in February 2022.

The primitive machine is a hotwater drilling (actually melting) installation from the early eighties, constructed at the time by AWI glaciologists. We were lucky to be able to have it work for us this year, since it could be its last deed with only 4 of 6 engines still working reliably. We were also lucky to have found two fantastic technicians willing to come with us and operate the machine, since there are only few people that know how to tame this monster device. Before the melting operation begins and the engines start howling, there is an excited tension in the air, all set, everyone ready? The holes we are melting allow us access to the water body underneath the ice shelf. Our first hydrophone is located 1 km from the ice shelf edge and will record the underwater soundscape of the surrounding open ocean.

At the same time, the protruding ice shelf shields the hydrophone from the intense iceberg traffic that passes in the adjacent coastal waters, that would otherwise damage our sensor. The hydrophone cable exits the ice to enter an aluminium box, which is packed with heavy car batteries to power the set-up, leaving a modest space for an acoustic recording module and amplifier that together register the hydrophone input to a series of SD cards. The box is buried in a snow shaft next to where the cable exits the ice and will be accessed only every three months by the overwinterers at the Neumayer III Base to exchange its SD cards and batteries. One kilometer to the south, another hydrophone cable is coming out of the ice, one km further south a third one. These two are our spare hydrophones that we also installed this season. They are not connected and still ‘dormant’, serving as the follow-up sensors once the distance to the ice shelf edge becomes critical. As the ice shelf is actually the edge of a large glacier, the whole system is advancing with ca. 150 metres per year towards the ocean, where every once in a while, large pieces break off and drift off as icebergs. The aluminium box can be easily disconnected and moved to the next sensor when needed over time, so that hardware loss (and scientific waste) can be reduced to just the hydrophone. In the days following the melting, the holes slowly close around our cables with which our recordings gain quality. After a noisy start, the data seem to quieten and a familiar soundscape reappears; PALAOA is online again.

PALAOA stands for PerenniAL Acoustic Observatory in the Antarctic Ocean. We have been collecting data here since 2006 and it is one of the places with highest acoustic species diversity, with ten marine mammal species that are reliably detected during their own seasonal acoustic windows. Not only a scientific treasure but also of stunning acoustic beauty, it is my favorite place on earth and it moves me to now be able to listen to this familiar soundscape again. The soundscape is like a sound place where I have spent so much time, listening, counting, describing and categorizing the different sounds that it is made up of. Through the headphones provisionally connected to the hydrophone, we hear Ross seals, icebergs and glacier calving, the bristling of the sea ice melting and every now and then a low frequency sweep from an Antarctic blue whale. A far from home acoustic homecoming.

Many thanks to
Oliver Roempler, Lukas Wilke and Elke Burkhardt

COLLABORATION

New GreenHAB Project on Harmful Algae Blooms in Greenland

Harmful algal blooms (HABs) pose a significant threat to ecosystem function and services, which can be critical in countries with a very sea dependent way of living, like Greenland. Although HABs are traditionally associated with temperate and warm coastal regions, HAB-forming species have increasingly infiltrated the Arctic Ocean due to global warming, intensifying the risk in these areas.

The GreenHAB project is designed to improve our understanding and prediction of indicators, tipping points and impacts of HAB species in the Arctic. We aim to achieve this by: a) characterizing current microalgal communities, with a focus on potentially harmful species; b) performing experiments to understand the ecological and evolutionary dynamics between harmful vs. non-harmful taxa under shifting environmental conditions; c) creating a dynamical ecosystem model to identify the critical environmental conditions and feedback loops leading to increased dominance of harmful microalgae (tipping).

Having a strong scientific foundation on HABs, the project places emphasis on engaging in dialogue with local and indigenous communities. This dialogue ensures the meaningful exchange of knowledge and invites collaborative development of research questions. Listening to local needs and practices is essential for creating operational measures to mitigate the impacts of HABs.

The GreenHAB project is part of the project network „Polar Regions in Transition“ funded by the Federal Ministry of Education and Research. It is coordinated by AWI, HIFMB is responsible for the dialogue with stakeholders.

Selected Recent Publications

+ more on Google Scholar: https://scholar.google.com/citations?user=uCoLTyAAAAAJ&hl=en

HIFMB TEAM

Fun Fact*

Spring is awakening! What does spring awake in you?