The 3D lab

A robot as co-author

Posted on January 19, 2023 by lembrechtsjonas

Ok, I’ll bite. The internet is literally full of fancy people jumping on the bandwagon called ‘ChatGPT’. The amount of lists I’ve seen floating by with ‘you are only using 1% of ChatGPTs full potential’ or ‘this is how I convinced the queen to marry me by using ChatGPT and so can you’ cannot be counted. But the thing is: we HAVE to talk about it. As this thing is a game changer (whether a big or a small game changer I leave open to discussion).

The potential applications for ChatGPT are numerous, but – let’s be honets – it’s important to acknowledge that it’s not as good as it looks for many questions you could ask it. Despite the opportunities it presents, there are also risks to consider. I have gained *some* insight on the tool and its appropriate usage now, and I wanted to share my perspective as an ecologist and lab leader.

First of all, it is important to note – as many have done before – that ChatGPT should not be used to write academic papers. Despite its ability to generate coherent introductions, deeper analysis of the text reveals a lack of credibility and depth. The tool lacks the capability to cite credible sources and often creates made-up papers as a result. It is thus unlikely that a thesis written by the robot would pass scrutiny: the introduction might sound good enough, but for the methods and results, and any depth in the discussion, they are still entirely on their own.

I know some of these people, but never ever have I published in Enviornmental Modelling & Software. So Veraverbeke, Bauwens & Muys, time to get writing!

Despite its limitations, ChatGPT can be a powerful tool when used correctly. Here are a few ways I have found it useful in my work:

Improve first drafts. Using ChatGPT to improve first drafts can be an effective way to save time and improve the overall quality of writing. By quickly jotting down ideas and creating a rough storyline, I can then feed this rough text through ChatGPT to improve sentence structure and coherence. This added step allows me to focus on refining and polishing the content, rather than spending time on correcting grammar and sentence flow. The ability to quickly identify and incorporate the good suggestions while discarding the bad ones, has significantly reduced the time and effort required to finalize my writing. Additionally, this approach ensures that the grammar and flow of sentences are already on point when I start polishing things myself, which ultimately leads to a polished and well-written final draft.
Ask it to change the tone of text. Why settle for plain and boring text when you can jazz it up with a little help from ChatGPT? By using it to change the tone of your text, you can add some pizzazz to your blog posts and tweets. ChatGPT can even serve as your personal brainstorming buddy, suggesting intriguing snippets of text and clever wordplay to make your writing more attractive. And while I can’t say I use the final version it proposes, I do like to sneakily steal a few improvements here and there. So next time you’re struggling to add a touch of personality to your writing, don’t be afraid to turn to ChatGPT for a little inspiration. (Ok, I admit, all the witty wordplay is entirely ChatGPTs-doing, this paragraph was significantly more boring!)
Shortening texts: I asked ChatGPT to suggest how to shorten a project proposal from 8 to 4 pages. It came up with great suggestions to regroup or rewrite sentences in a more compact way, which also helps with email writing and other types of writing that need to be concise.
Annotate R-code. This is a relatively new discovery for me, but I think it will be a GREAT tool for my students! I just feed in a piece of R-code and ask it to annotate it, and it writes down in text what all the steps mean. For students, who are often unfamiliar with many of the tips and tricks of R that I learned over the years, this can make R-code I sent them a lot easier to understand.
Suggesting formatting for things I am not familiar with writing: For example, I haven’t written too many recommendation letters yet, so I ask ChatGPT for an example by feeding it some keywords for the candidate in question. This gets rid of the ’empty page’-issue where you don’t know how to start writing and lets you build on a good foundation.

My final words are perhaps the most important part of this blogpost: we HAVE to discuss this tool with our students. They are more tech-savvy then we are, so they will find out about it. Yet they might not get all the pro’s and cons, and they might especially brush over the ethics. Can they use it? Sure, I’m all in favour of getting all the help one can, the main goal of their work (in their master or PhD thesis, mind you, not in other courses) in my opinion is anyway to advance science and grow as a human in the process. They can use whatever tool they want, be it a fancy and expensive measurement tool or a chatty robot. As long as they know the strengths and weaknesses. That’s the same for that fancy measurement tool for that matter.

It’s a common question – is the text I’ve written still mine when I’ve utilized a language model like ChatGPT? I firmly believe it is. While the model may offer suggestions, I ultimately have the final say on what to include or not. Think of it as having a co-author, one who carefully reads and offers suggestions, but it’s ultimately up to me to decide what makes the final cut.

Posted in Science | Tagged Academic writing, AI, Artificial Intelligence, ChatGPT, Ecology, Science, Writing | Leave a comment

Database progress

Posted on January 16, 2023 by lembrechtsjonas

Last week, we bid farewell to Amber, a bittersweet goodbye that marks the end of a successful chapter in our ‘SoilTemp’ project. Yet it also provides closure to a very good story, as you’ll see.

Amber joined us as a master’s student and wrote an outstanding thesis on the air conditioning effect of nature in gardens, in the framework of our ‘CurieuzeNeuzen’ citizen science project. She then joined us part-time as a database manager for the SoilTemp database, and our achievements together during that time were nothing short of fabulous.

So how did we get Amber on board? After getting a few grant rejections, I sent out a tweet in desperation, warning the community that our planned work with SoilTemp was in jeopardy. But as fate would have it, the community rallied around us! GEO Mountains, an initiative focused on global environmental monitoring in mountains, for example, offered a concrete solution that resulted in this part-time job for Amber. Our main goals were two-fold: to make significant strides towards open access publication of the SoilTemp database and to increase its coverage in mountain regions worldwide, particularly in those hard-to-reach areas where even weather stations are scarce.

Sensor installed in my own favourite cold-climate mountain region in the northern Scandes, where it monitors birch forest understory temperature in a dense field of *Empetrum* and *Vaccinium* heathland

Amber sailed through the backlog of data submissions like a database diva and reprocessed all 35,000 time series in a format that will allow them to be entered into our upcoming relational database. The latter is being accomplished together with Rémy from iDiv (Germany), another epic story for another day. And our second goal? A mountain of success. We equipped a whole lot of regions participating in the Mountain Invasion Research Network (MIREN) with sensors, and we’re seeing increasing monitoring in many cold-climate mountains, tropical mountains, and especially African regions. The large white gaps on our map are shrinking fast!

So, thanks Amber, and thanks GEO-Mountains, we definitely made the best of it!

Posted in Science | Tagged Database, Ecology, Microclimate, Mountains, SoilTemp, work | Leave a comment

Battling my emails

Posted on January 3, 2023 by lembrechtsjonas

I started tracking my time at work in detail at the start of my postdoc in 2018, using the amazing app ‘Timeular’. This series of stories provides some insights into postdoc life using that data.

A traumatizing aspect of academia to many post-PhD scientists, perhaps: the inbox! Throughout my postdoc career, I noticed soon enough that the tide of emails was swelling, especially from the moment – in 2019 – that I put myself at the steering wheel of SoilTemp, a network of several hundred scientists. An increasingly international network with increasingly many collaborations, and you can imagine how many emails would come in! Yet, how much of my time was actually going into emailing?

Numbers, please!

Percentage of my time spent emailing since the start of my postdoc in 2018, tracked minutiously using the Timeular tracking software.

At the start of my postdoc, the mailbox (and its fancy relatives like Slack, Microsoft Teams, or even Facebook Messenger for some colleagues) only ate up a blissful 7% of my time. By 2020, with SoilTemp in full swing, I rounded the magical number of 20%. In a 5-day week, that’s one full day of emailing… Daunting! And no wonder I started to consider my efficacy.

If that amount of emailing is truly problematic, that’s a different story, of course. Not only was I managing this large network with its seas of useful and important emails, a lot of my other tasks were also increasingly replaced by an advisory role. Statistical analyses and paper writing as done in my PhD and at the start of my postdoc was now turning into assisting students or colleagues to do just that. Many of their questions could often be answered in a short – or longer – email, and them taking over lead authorship on many papers implied that my own time spent on these papers was reducing.

Nevertheless, I was realizing that I needed to be smart and efficient to keep that tide of emails at bay. I decided to step away from the immediate answering that I was used to from back in the days and delineated specific moments in which I would email. Now, I’m starting my Monday, Wednesday and Friday by working through my inbox and putting it back to zero. I answer emails that only take me up to 10 minutes of work, and schedule more demanding ones for different times, depending on their deadline.

This way, I am usually relatively fast with my response (2 days maximum), while still keeping a significant chunk of my week email-free. I do make an exception for my students, of course: their questions, thoughts, musings, or whatever they feel the need to share, will usually get a much quicker, if not immediate, response (as long as I can answer in less than 5 minutes).

Is this strategy paying off? I bet you it is! I regained some sanity, I am still on top of my inbox, and I turned the tide of emails: in 2022, the percentage was down to 14.5%, a more than 5% drop since 2020! Very happy the numbers support that decision!

Microsoft Teams – and the IT support team asleep next to it

Posted in Belgium | Tagged Cats, Mailing, Science, Time tracking, Timeular, work, Work efficiency | 1 Comment

Hitchhiking mountain roads and trails

Posted on December 29, 2022 by lembrechtsjonas

The Mountain Invasion Research Network (MIREN) has made itself ‘world famous’ through its protocol for the long-term monitoring of vegetation along mountain roads. We even got a recent publication dedicated in its entirety to that protocol! However, mountains – and the rest of our increasingly connected world, for that matter – are crisscrossed by a whole bunch of other ‘linear disturbances’. An obvious question thus arose: are the vegetation patterns we see along roads mirrored along these other linear disturbances, such as trails?

How problematic to nature are mountain trails – linear disturbances that protrude deep into our most precious protected nature.

To answer that question, we set up a second global vegetation monitoring protocol, yet now focussed on mountain trailsides. While our roadside protocol has been adopted in more than twenty regions worldwide, the tally for the trail protocol is currently at six. So a tad more humble in its scope, yet still more than enough to satisfactorily answer a first important general question – at the core of our MIREN business: are patterns in non-native species richness along mountain trails a mirror of those found along mountain roads?

Hiking along mountain trails can bring non-native species to surprisingly high elevations. In this stony desert above the treeline in the Chilean Andes, we found some European Taraxacum flowers!

The answer can be found in a recent publication – part of a new book on the role of tourism and recreation as drivers of biological invasions.

So how do our trails compare? Observed levels of plant invasion were substantially lower along trails than along roads, adding up to only about 20% of the species numbers. Nevertheless, the same decline in richness with elevation is found that is so characteristic of mountain roadsides across the globe, and many of the most common non-native plant species along roads are also found along trails.

Most striking, however, is the lack of decline in non-native species richness with distance to the disturbance. Along roads, we see worldwide the same characteristic pattern, with high numbers of non-natives in roadsides and then a more or less steep drop in species close to the road in the interior vegetation. Along trails, this pattern is virtually not there, which implies that levels of invasion are largely the same in trailsides and the interior vegetation.

Non-native species richness along mountain trails (red line) and in the adjacent natural vegetation (turquoise line) across six mountain regions (see map). Yellow-to-brown ‘smudges’ on the map depict the global mountain regions.

The latter has some worrisome implications, as it shows that the resistance to invasion of the interior vegetation close to trails is much lower than along roads. This could be due to the sparser vegetation (most studied trails were at higher elevations in the mountains than the roads), but most likely also to the differences in disturbance dynamics: while roadsides are often highly disturbed close to the road, trailsides are less disturbed in general, yet the ‘disturbers’ often wander off the trail more, spreading its impact further into the interior vegetation.

European dandelion in the alpine zone of the Chilean Andes

The results of our first global comparative analysis of roads and trails are thus both encouraging and worrying. Encouraging for our trailside managers is the fact that levels of invasion are lower here than along roads. Nevertheless, the lower resistance against invasion of the surroundings asks for strong preventive action to limit the impact of invasions along mountain trails, especially those in vulnerable protected areas.

Reference

Barros et al. (2022). The Role of Roads and Trails for Facilitating Mountain Plant Invasions. https://doi.org/10.1079/9781800620544.000. Full text here!

Posted in Science | Tagged Ecology, Global change, Hiking, Mountains, Nature photography, Plant invasions, Science, Trails | Leave a comment

Long-term reporting

Posted on December 23, 2022 by lembrechtsjonas

A big chunk of my work focuses on long-term monitoring, of vegetation and (micro)climate. It highlights the importance of this continuity, and many of the most important conclusions we are taking – and especially will take in the near future – depend on this persistence in the field.

Yet persistence is far from easy. Repeating your fieldwork year after year, finding the necessary funds, time, people and expertise to keep going, it is time and again a challenge. It is also very hard to plan for, as who knows where one will be five years from now? Especially in the highly volatile conditions of an early career in science in the 21st century (but that’s a whole different story).

And yet, we persist. What also persisted, is this website. So dear to me since the very start of my PhD, I managed to keep going the stories here year after year, despite the waining interest in ‘macroblogs’ in favour of microblog platforms like Twitter. What’s even better: my audience is steady. I saw a continuous increase in visitors throughout my PhD up till 2018, when I defended. Then, as time availability for writing reduced, yet content got perhaps more interesting (I was publishing more papers, for one), the audience has kept popping by.

Yearly visitors to http://www.the3dlab.org, formerly called lembrechtsjonas.wordpress.com

In one of the first posts I wrote on this blog, I warned the reader that climate and land use change were awakening a sleeping dragon in the Arctic.

It is my job to warn the world for this sleeping dragon.
It is my job to find a way to preserve his night’s rest.
It is my job to predict his next move, so we are not unexpectedly catched by his swirling tail.

I still find that beautiful. And I do believe that 2013-me would be very proud of the job I did since then.

Stay tuned in 2023, for more!

Majestic view over the mountains in northern Norway, the land of the sleeping dragon.

Posted in General | Tagged Blog, Mountains, Nature, PhD, Science | Leave a comment

Microclimate symposium

Posted on December 13, 2022 by lembrechtsjonas

Truth be told, we just had a little pot of money that needed spending. That pot of money came from a successful application to the University of Antwerp to enhance communication between universities.

Now, communication between universities, that’s my kind of thing, and we realized we had just the set-up at hand: the high density of microclimate research groups in Belgium and beyond. A quick invitation to our friends in Leuven, Ghent, Liège and Amiens (also basically Flanders, no?) immediately gave a positive response: we all agreed that it would be great to sit together for a few days and learn from each others successes, deep dives and mistakes.

To make things even more attractive, we lured in some big names from abroad, and soon we had a workshop programme buzzying with potential. This, combined with a perfect location and host lab (the ForNaLab at Ghent University, embedded in its own research forest, and the workshop was destined to be successful.

Outdoor forest heating experiment – basically just two steps from the offices of the lucky ForNaLab-scientists!

And I know, we had a fantastic microclimate conference only recently, but we felt there was room for something else: a smaller group of people, with more room to go in-depth on the tricky stuff. Especially the younger members of the labs could benefit from a bit more time to explain their work, their goals and dreams, and the things they struggle with. Together, we might have some answers that could propel their work forward.

The research tower in the forest is positioned just in between a few different species of trees, for optimal monitoring.

So that’s how we got to spend two days in a forested environment outside of Ghent, and how I got another notebook full of great ideas and suggestions for the future.