From Prototype to Reality: Moving Beyond the Technology Hype in Ecological Research

Abstract

In the face of accelerating anthropogenic pressures threatening global ecosystems, the need for environmental monitoring grows ever more important. Continuous tracking of species and ecosystems helps us to understand ecosystem dynamics and functioning in these uncertain times. Scientists and conservationists everywhere are emphasizing that technological advancements can improve monitoring efforts by increasing spatial and temporal resolution and allowing for real-time data streams (Hahn et al. 2022, Speaker et al. 2022). The enhanced understanding of global ecosystem responses to, for instance, climate change is critical to inform policy decisions and guide conservation and environmental management (Allan et al. 2018). More and more, the scientific and conservation community are creating new affordable monitoring systems fully tailored to their specific needs, mostly by leveraging open-source electronics such as the Arduino or Raspberry Pi platforms (e.g., Pearce 2012, Jolles 2021, Mühlbauer et al. 2023). These success stories are published in various scientific journals, but when problems or failures occur they remain underexposed, potentially creating inflated expectations for readers (Lahoz-Monfort et al. 2019). While reflective practice (see Box 1) is strongly embedded in social sciences research, it is rarely adopted in the natural sciences, let alone distributed via scientific publications (but see the Centre and Journal of Trial and Error; https://trialanderror.org/).

This piece of reflection grew from the complications we faced and insights we gained when developing an innovative phenology monitoring system ourselves. With this writing, we want to look beyond the technology hype and associated good news narratives to think about a future with sustainable technological innovations for ecological research and conservation. Sustainability, in this case, implies performance, stability, and reuse, which is definitely not the same as achieving successful measurements over 1 week, month, or year for one project. We firmly believe in the potential of technology to help us better understand the challenges our planet is faced with and we hope this testimony can serve as a motivation to take on ambitious innovation projects but with a realist view and a collaborative mindset.

Within our research project, we study how global environmental change affects flowering and vegetative phenology of forest understorey plant species. More specifically, we assess phenological shifts in a global change mesocosm experiment in which small plant communities are being exposed to warming, light addition, and nutrient addition. To investigate flowering phenology, we counted flowers every 2 days throughout the flowering season of 2021 and 2022 (Lorer et al. 2024). Since the temperate forest understory harbors both early flowering and summer flowering species, this meant starting in February and ending in October. To monitor leaf phenology and intra-annual height dynamics of the plants, we measured plant height and cover for one whole year in 2023. Summed together, more than 900 person-hours (i.e., 120 full days) were spent for these field observations in only one location. Going into the field and observing the study species daily to weekly is indeed still the most conventional way to study plant phenology, but is time-consuming. This perfectly illustrates how important automated monitoring devices could be for phenology and plant growth studies. Different phenology camera approaches already exist, such as Phenocam (https://phenocam.nau.edu/webcam/) and Wingscapes Plantcam (Laskin et al. 2019), but these are not customizable and come at high costs. And so, enthused by promising examples in the scientific literature, we started to work on a new cost-effective automated monitoring device.

We conceptualized a phenology sensor, equipped with multiple environmental sensors to monitor plant growth and the microclimate around the plants simultaneously. The device builds on the MIRRA (Microclimate Instrument for Real-Time Remote Applications) microclimate monitoring system (Pieters et al. 2021) which we customized (both hard- and software) and expanded with a stereo-imaging component. MIRRA is a networked system of modular nodes, each supporting multiple sensors to measure several microclimate variables: soil and air temperature, relative humidity, and light intensity (Fig. 1). Nodes, custom-built printed-circuit boards (PCBs), periodically send their microclimate data to a gateway which uploads the data to a central cloud server, accessed by users through a web portal. The imaging component consists of two ESP32-CAMs, which are low-cost microcontrollers equipped with a 2 MP image sensor and a microSD card reader for local storage of the images (Fig. 1). The device is programmed to take pictures at sub-daily time steps and record microclimate data at the same time. Having two cameras suspended above the forest floor allows to characterize intra-annual plant height dynamics, that is, the phenology of vegetative growth, using close-range stereophotogrammetry techniques.

While we succeeded to work out an operational device that met many of our intended goals, we were not entirely satisfied with the end product. Moreover, we encountered many unforeseen obstacles along the development process, making it much more time-consuming than anticipated. This is indeed a common issue in the ecology and conservation community, yet often overlooked (Hahn et al. 2022, Speaker et al. 2022).

As ecologists without training or experience, but a great interest in technology and product design, we started conceptualizing our device by identifying the needs based on our plant ecological knowledge. Due to a lack of in-depth technical knowledge, we initially did not completely grasp the technical complexity of these needs. Really understanding your device's needs is often disregarded but is crucial because the choices made during the first conceptualization steps define the direction of the following design process (see Fig. 2). Furthermore, during this first step it is essential to ask yourself whether you are introducing a new device because of its novelty or because it will genuinely solve your problem. It is easy to be captivated by the idea of creating cutting-edge technology and forget about making these tools actually performant, stable, and user-friendly.

In the next step, we browsed for proof-of-concept projects in the scientific literature on ecological and meteorological monitoring devices. We found many but chose to adopt the MIRRA system mentioned above. We also consulted many online blogs and maker tutorials to learn about the involved hardware and software. These information sources helped the conceptualization of the envisaged device but we remained uninformed about the product design process that would have to follow. As a result, we did not formally translate our needs into specific features for the monitoring device; instead, this unfolded gradually throughout the process. This approach impeded us from verifying the feasibility and compatibility of every feature so were inevitably confronted with several issues which could have been anticipated with a formal feature list evaluation and optimization step.

Moving on, we quickly got stuck between the “Build prototype” and “Test” phases. While many projects we found in our literature search started deploying devices that were still in the prototype phase, data quality was very important for our application so this was impossible in our case. We kept revising our prototype within the confines of the existing design (e.g., in terms of chosen components) without exploring real alternatives to address certain limitations. Yet, sometimes a decent update entails a complete redesigning or rethinking of the problem. Sometimes you need to let go to go forward.

Finally, it cannot be stressed enough that a sufficiently long testing phase should be built into the project plan. Comprehensive testing means testing the device in various field conditions, over long periods of time to ensure long-term performance and stability. The length of the test phase and the chosen test strategy depends on the complexity of the developed tool and the required data quality, but scaling and deploying should never be rushed.

While we now understand that we are definitely not the only ones struggling, we believe that we would have adopted a more sensible approach from the onset if stories similar to ours would also have an outlet or if success stories would encompass the integral development process, including failed attempts. Most technological advances in ecology and conservation stem from uncoordinated inconsistent initiatives often leading to duplication of efforts and, in many cases, incomplete development processes (Joppa 2015; Lahoz-Monfort et al. 2019; Speaker et al. 2022). Obviously, this “wastes time, money and resources in a discipline that can ill-afford to do so” (Joppa 2015). So, where do we head from here?

First of all, not all technological innovations require complicated solutions. There are plenty of examples of well-functioning simple monitoring devices designed by ecologists or conservationists. On the other hand, producing advanced purpose-built technologies, like ours, requires a set of skills beyond the capabilities of one person so forming cross-disciplinary collaborations is crucial (Schulz et al. 2023), something we only realized during our design process. We involved electrical engineers at our university on an ad hoc basis to help us with our further device development, but challenges persisted. We think that with an interdisciplinary collaboration from the outset, with a formal involvement of electrical and computer engineers, product design experts, and ecologists, we would now find ourselves with a more advanced product. In a true interdisciplinary collaboration (going beyond a multidisciplinary approach) ideas and perspectives are integrated across disciplines to solve one common research goal with new knowledge for all involved disciplines. The importance of interdisciplinarity for technological innovation in ecology and conservation has already been pointed out repeatedly over the last decade (Joppa 2015; Allan et al. 2018; Lahoz-Monfort et al. 2019; Besson et al. 2022; Hahn et al. 2022; Schulz et al. 2023). Involving engineers and computer science experts would call their attention to the urgent need for novel technologies to advance our understanding of ecosystems in the face of global change, prompting them to effectively and proactively use their expertise to protect biodiversity and ecosystem services. Yet, examples of true interdisciplinary collaborations remain rare.

For some applications, it can be of interest to include end-users in the development process. In short surveys, end-users can communicate their needs for measuring devices, concerning ease of use and deployment, desired accuracy, and battery life and capability under harsh conditions (Speaker et al. 2022). Doing so can generate an impact beyond a single research question. For instance, connecting with potential users can facilitate the scaling up of the sensor network to cover larger geographic ranges, and studying a device's relevance for research or conservation purposes other than your own, may allow you to enhance its built-in versatility.

Achieving true interdisciplinary collaborations comes with many challenges, such as communication barriers and difficulties with the integration of perspectives. In order to stimulate such future collaborations, an interdisciplinary mindset and competence should be cultivated within our academic and educational system (as exemplified by the Centre for Unusual Collaborations; https://unusualcollaborations.ewuu.nl/). Cross-disciplinary collaborations can be translated to educational exercises where students in the biological sciences work together with engineering students to jointly solve an ecological research question by inventing novel technologies. During these group projects biology students are exposed to extensive computational thinking and technology design and manufacturing processes. Electrical engineering students learn about the ins and outs of ecological research and nature conservation and develop a raised awareness toward the impact of global changes on ecosystems. They can discover how to use their knowledge to contribute to these fields, bringing more societal relevance to their work. Moreover, designing new monitoring devices comes with many engineering challenges (Joppa 2015). Devices need to be strong, light, power efficient, easy to deploy, multifunctional, but still cost efficient. In other words, a host of requirements come together posing an intriguing challenge to tomorrow's engineering students potentially elevating their interest in this domain.

Working on this project, we experienced first-hand the iterative nature of product design, which follows a spiral development process requiring multiple build, test, fail, and revise (and upgrade) iterations (see our interpretation of this process in Fig. 2). Failure is thus inherent to this process in order to improve the end product. However, our project funding (just like most short-term funding) was not matched to this kind of cyclical time investment, making it very hard to build sustainable products for long-term applications. Unsurprisingly, unsustainable financing is indicated as the principal constraint for technological innovation in this domain (Speaker et al. 2022). And, the competition for scarce incompatible funding leads to overpromising in project proposals to secure financial support, often with disappointing outputs at the end of the grant period.

In conclusion, we believe that for a sustainable future of technological innovations for ecological research and conservation, we need to (1) dare to disseminate difficulties and failures, (2) transcend disciplinary boundaries and (3) provide dedicated research funding to support the iterative process of product design and facilitate interdisciplinary collaborations.