Little resistance where we expected it

Posted on September 22, 2026 by lembrechtsjonas

Once you have gone through the effort of collecting a global database, you can start playing with it. And one of the most fun things to play with, I’d say, is general ecological theories – the kind that you would think hold everywhere, but that usually play out differently depending on where and how you test them. We’ve done a few of those with our MIREN plant survey database from along mountain roads, and here’s another important one, just published in the Journal of Vegetation Science (Buhaly et al. 2026).

This time, the theory on the chopping block is biotic resistance: the old and intuitive idea, going back to Elton in 1958, that a diverse, well-established native community should be harder for a newcomer to break into. More species already occupying the available niches, competing for light, water and space, should mean fewer open doors for a non-native plant trying to move in. It’s one of the most-cited explanations in invasion biology, but it rests on the assumption that competition is what dominates plant-plant interactions. And we’ve known for a while that this assumption gets shakier the harsher the environment gets: in stressful places, facilitation (plants actually helping each other, via shelter, warmth or moisture) tends to take over from competition. High-elevation, alpine environments are about as stressful as it gets. So does biotic resistance survive the trip up to the treeline and beyond?

The scenery of this paper – one of our Norwegian mountain roads

Testing an old idea

Together with the whole consortium, and led by Meike Buhaly, we used our MIREN road survey data to find out, using 15 mountain regions on six continents. At each site, a plot runs along the roadside, and a second one starts 50 m off the road and heads into the (semi-)natural vegetation behind it, the same paired design we’ve used since 2007, and that has featured oftentimes on this blog here. On top of that, we (well, mostly the lead author team in Germany, I just collected the leaves) built a proper trait dataset, with nearly 1,500 leaf samples, run through a near-infrared spectrometer back in the lab, to calculate not just how many native species a plot holds, but how functionally different those species actually are from one another.

Spelling out our expectations for this paper, based on conflicting theories: in Hypothesis 1 (H1), we expected that at low elevations, fewer non-native species would be found in communities with high native diversity due to competition. We hypothesized the opposite at high elevations (H2), with diverse native communities facilitating the establishment of non-native species. In Hypothesis 3 (H3), we anticipated a steeper slope of the relationship between non-native species richness and native diversity in natural habitats. Additionally, in Hypothesis 4 (H4), we hypothesized that the ratio of non-native species in natural versus disturbed roadside habitats would be higher in high-elevation environments than at low.

No resistance from richness – sometimes the opposite

The graph supporting the conclusions here. A bit counterintuitive: the y-axis shows the slope of the relationship between natives and non-natives. Positive values suggest more non-natives in species rich plots, which we unexpectedly see happening at low elevations (top panel). At high elevation, the relationship is neutral rather than positive.
For functional diversity (bottom panel), the pattern is very different

The short version: using native species richness as the classic proxy for biotic resistance, the expected resistance pattern didn’t show up anywhere in this dataset. If anything, it was reversed at low-to-mid elevations, where non-native richness was consistently higher with native richness. This was true, unexpectedly, both at the roadside and in the natural vegetation, plausibly because species-rich, structurally complex lowland communities simply offer more niches and microsites, for natives and non-natives alike, than their species-poor counterparts.

Looking at functional diversity told a different story, however. At low elevations, functionally diverse natural communities did host fewer non-natives – that’s one piece of actual biotic resistance we did find! But climb the elevation gradient, and that relationship weakens, then flips: at the highest elevations, functionally diverse native communities host more non-native species again. Exactly what you’d expect if facilitation, not competition, is running the show up there, as has been shown before for native cushion plants acting as “nurse plants” for exotic invaders in the high Andes.

So: some evidence for biotic resistance at low elevations, when using functional diversity, but overall a fairly strong case for the opposite to be happening.

Going back to Norway 2012

Then, at the end of the paper, there’s a part that I am a big fan of personally. Back in 2012, as a master’s student doing the first MIREN survey in northern Norway, I found that the ratio of non-native to native species in the roadside – compared to the untouched vegetation right behind it – increased steeply with elevation. My reading at the time: weaker resistance up high lets a disproportionate share of whatever non-natives do arrive spill off the road into the natural vegetation. That finding did get published, in my first ever peer-reviewed publication, but I always felt a bit hesitant. I trusted the pattern and process, but I felt like we just didn’t have enough data – very low invasion levels in Norway – to confidentially accept the claim as proven.

The global dataset now shows that this ratio does indeed climb toward high elevations, as we saw in Norway. However, it’s not a simple upward slope. It’s a U: the ratio of non-natives in natural vegetation relative to the roadside is highest at both ends of the elevation gradient (around 0.6 at each extreme), and dips in the middle. The high-elevation arm of that U is the one I stumbled onto in Norway — non-natives that make it up the road managing to spread into the surrounding alpine vegetation surprisingly easily, either helped along by exactly the facilitation signal described above, or by an overall reduction in interactions (my reading). The low-elevation arm is a different story, though: lowland sites are where most non-natives arrive in the first place, have had by far the longest time to build up a presence, and can draw on a far bigger regional species pool, so plenty of them eventually work their way off the road there too, just through sheer numbers and time, and regardless of the resistance levels.

The famous U-shaped graph of the ratio of non-native species in the natural vegetation versus the roadside

Which also explains, rather neatly, why my Norwegian roads never showed that low-elevation bump: Norway’s cold climate, isolation and modest regional species pool have always made it something of an outlier when it comes to that kind of long-term, low-elevation invasion build-up. What I picked up on in 2012 was thus really just one half of a much bigger, two-sided global pattern, and it took fourteen years and a lot of collaborating mountains to see the other half. Glad!

The native Angelica archangelica in a Norwegian roadside

Reference: Buhaly, M., Turner, S.C., Kreuz, N.K., Vandvik, V., Veltmann, B., Alexander, J.M., Lembrechts, J.J., et al. (2026). No signal of biotic resistance to non-native species establishment in high-elevation communities. Journal of Vegetation Science, 37, e70174. https://doi.org/10.1111/jvs.70174

Posted in Science | Tagged , , , , | Leave a comment

Proving extinction is hard. But we can do so now.

Posted on September 3, 2026 by lembrechtsjonas

For decades, climate change ecology has mostly been a story about shifts: species moving upslope, poleward, tracking the climate they’re adapted to as it slides out from under them. It’s one of the best-documented fingerprints of warming in nature.

What we’ve been much worse at showing is what happens at the far end of these shifts: actual extinction. Not “the range is contracting”, but the species being gone, locally, for good. That’s surprisingly hard to prove, and that for two good reasons. First, a single before-and-after comparison – the kind most historical resurvey studies rely on – have a hard time telling a real decline apart from population fluctuations or missed observations; you need repeated resurveys over time, at the same place, to separate signal from noise. Such standardized long-term surveys are rare. Second, alpine specialists are tough, long-lived plants that respond to a changing climate with a considerable lag, so a species can be quietly suffering in silence for decades before it actually disappears. So even though we’ve had every reason to expect that warming is pushing cold-adapted specialists off the top of mountains, actually catching that happening – at scale, and in relation to how much a place has warmed – has remained tricky.

In a new paper out this week in Science, driven by Johannes Wessely, Stefan Dullinger, and a long list of GLORIA colleagues, we finally got to prove what everyone already feared.

Where we looked for extinctions. The map shows the 16 mountain regions in our study, with circle size showing how much local extinction we found there – bigger circles, more species lost. On each of the 62 summits, researchers return to the same 16 permanent 1×1 m plots every seven years, arranged just below the peak (right), so we’re always comparing exactly the same patch of ground over time.

The reason that I – as an ecologists from the ‘lowlands’ – am so drawn to the mountains for my research, is that these mountains are close as we get to a clean experiment. The mountain summits we looked at here experience very little land-use change, no fertilization, none of the usual confounders that make biodiversity loss so hard to pin on any one cause. The GLORIA network has resurveyed the same 1×1 m plots on 62 summits across 16 European ranges every seven years since 2001, resulting in four surveys, 21 years of data from 896 permanent plots. If a species disappears from a mountaintop, warming is one of the few plausible suspects left.

High-elevation plants, like this Gentiana, are increasingly under pressure

Summit plant diversity has been rising for years, as lowland species colonize newly hospitable ground. We knew that, and that obviously remains true. But behind that increasing richness, local extinctions have been rising too. What’s more: these extinctions are getting faster with each survey, and this at every scale we checked: plots, summits, whole mountain ranges. Richness went up because colonization outpaced extinction, but that fact was hiding the losses. And what we are loosing, unfortunately, are exactly the cold-adapted specialists that make high mountains botanically unique.

Extinctions on the rise! Each dot is one mountain summit, showing what share of its species disappeared between one survey and the next. Color shows how much that summit had warmed till then – yellow is more warming, purple is less. Two things are important to note here: dots (and the trendline) clearly get higher over the three survey periods, and the warmest summits tend to sit toward the top.

Importantly, what predicted these extinctions was more than just time. Where the paper gets scarily elegant is where it shows that these extinctions truly correlate with how much a summit had actually warmed. Every extra 0.1°C over the preceding 14 years came with a measurable rise in extinction rate, and warming actually explained far more of the pattern than time alone. Species were also more likely to disappear from their warm, low-elevation range margin, and from communities shifting hardest toward warmth-loving species. Good to know, perhaps, as well: a preceding decline in cover was – while noisy – an early warning signal of extinction.

An extinction debt coming due?

Alpine specialists are tough, slow-growing, long-lived and thus often lagging behind the changing climate. That raises an uncomfortable possibility: what we’re seeing may be the first instalment of an “extinction debt” built up over decades of past warming. Now, that extinction debt is starting to be paid, which means that more losses could already be written in stone.

We’re not claiming climate change is the biggest driver of biodiversity loss everywhere; land-use change and nitrogen deposition still dwarf it in most human-dominated landscapes. But on summits, where those confounders barely exist, the fingerprint of warming on real local extinctions is now about as clear as it gets. And it’s growing. Measurably.

Reference: Wessely, J., et al. (2026). Rising plant extinction rates on European mountain summits. Science.

Posted in General | Tagged , , , , , , | Leave a comment

How much hydrology hides in our microclimate sensors?

Posted on August 27, 2026 by lembrechtsjonas

I’ve spent over a decade now sticking small microclimate loggers into the ground, mostly to talk about temperature. Soil moisture has always been the awkward third wheel in that story: TOMST loggers measure it too, our CurieuzeNeuzen in de Tuin citizen-science project collected it by the thousands, and yet I’ll be honest – until recently we had done remarkably little with it. Soil temperature is intuitive, comparable across studies, and (relatively) forgiving. Soil moisture is none of those things. It depends on soil type, on calibration, on what exactly you mean by “moisture” in the first place, is extremely variable and rarely statistically ‘normal’. So it sat there in our growing database, a bit unloved and ignored.

CurieuzeNeuzen in de Tuin citizen scientist collecting a soil sample, next to their TOMST TMS-NB ‘garden dagger’

That changed, fortunately, when the hydrologists at the Vrije Universiteit Brussel came knocking with a very interesting question: can our garden sensors tell us anything about groundwater?

Borrowing eyes from a different field

The premise of the collaboration (now published here!) was nicely elegant. They had already built and run mHM, a proper physically-based hydrological model, for the whole of Flanders. The kind of model that needs detailed climate forcing, soil maps, land use, and a fair chunk of computing time, and that spits out things one actually cares about for water management: groundwater recharge, evapotranspiration, degree of saturation. Variables for which you cannot simply stick a sensor into the ground and read off directly at the same low cost we do for temperature and soil moisture.

Meanwhile, we had thousands of CurieuzeNeuzen in de Tuin loggers with two summers of near-continuous, low-cost, noisy, wonderfully abundant soil moisture and temperature readings, measured in places a hydrological model would never normally get near: backyards, parks, the occasional forgotten corner behind company head quarters.

We got thousands of temperature and soil moisture sensors, which have up till now been used only for half their potential.

The question the VUB team asked was basically: if we feed a machine learning model nothing but our sensor time series – no rainfall, no radiation, no coordinates, just the lags, rolling means and differences of what the sensor itself reports – can it learn to reproduce what the full hydrological model would have said for that spot? In other words, how much hydrology is already implicitly encoded in a humble soil moisture logger, once you look at it the right way?

Where it all happened: the coloured speckles are individual CurieuzeNeuzen in de Tuin sensors scattered across Flanders, with the three test areas from the paper zoomed in — the whole region, the Demer sub-basin, and the small independent test site in Boechout.

The answer: quite a lot, especially in bulk

Using LightGBM (a fast, tree-based algorithm, chosen after testing a handful of competitors), they could indeed emulate mHM’s estimates of groundwater recharge, evapotranspiration and degree of saturation reasonably well at most sites, capturing the difference between the miserably wet year of 2021 and the properly dry year of 2022, catching the seasonal rhythm, keeping bias low. Not perfect, and definitely not a replacement for the real model (the paper is quite upfront about that), but good enough to be genuinely useful as a diagnostic and extrapolation tool.

One of our better sites: our sensor-based machine learning model (green) tracking the “real” hydrological model’s groundwater recharge estimate (orange) almost peak for peak, in both a wet year (2021, left) and a dry one (2022, right). Not every garden did this well, but this is what it looks like when it worked out nicely!

The part I find most interesting, though, is the following: on their own, individual sensors are a weak signal for this kind of hydrology – a single logger in one garden tells you a bit about that garden, yet very little about regional groundwater dynamics. Not a surprise there, as we knew the soil moisture signal was messy and hard to trust. But once you start aggregating across many sensors, the picture sharpens considerably: the density experiment in the paper shows that as more sensors get pooled together, the bias in the estimate stabilizes and the noise from single quirky locations melts away. It is, in the most literal sense, a case of the network being worth more than the sum of its parts. That’s reassuring to see in print, as that was the point I had been making about these sensors all along!

Pool just a handful of sensors (left side of the graph) and the bias estimate varies wildly depending on which ones you happened to pick (the shaded bands are wide). Add more sensors moving to the right, and average them together, and that uncertainty collapses, even though the mean bias barely moves. One noisy garden sensor tells you little; thirty-five of them, averaged, predict the truth.

There’s a nice practical postscript too: when the model, trained purely on 2021–2022 Flanders data, was applied to an entirely independent site near Boechout (Flanders) the following year, its recharge estimates tracked the ups and downs of actually measured groundwater levels pretty well. That’s the kind of transferability test that is very welcome to see.

Why this matters (to me)

For me personally (selfishly), this paper did two important things. First, it took our soil moisture data out of the drawer we had secretly already put it in (“nice to have, hard to use”) and showed it has real value beyond microclimate ecology – groundwater managers, drought forecasters and agricultural planners could plausibly use dense, low-cost (citizen-science) microclimate networks like this one as a genuine complement to expensive physical models, especially in places where running a full hydrological model isn’t feasible.

Second, it was just refreshing to watch hydrologists do to our data exactly what we do with vegetation or temperature data: squeeze out patterns we hadn’t thought to look for. Another example of where microclimate measurements can inform other disciplines, as we have showed before for many other applications.

There is plenty left to do, of course, and the paper has a whole lists of suggestions: we had no winter data (when most European recharge actually happens), no proper soil calibration (remains a bottleneck for the TOMST sensor soil moisture data), no guarantee the same approach transfers to a different climate or a different model. But as a first demonstration that “yes, there’s real hydrological signal buried in a network of €100 loggers, if you have enough of them and someone willing to dig it out” – I’ll take it!

Reference: Elsaidy, A., Lekarkar, K., Yimer, E.A., Van de Vondel, S., Lembrechts, J.J., Meysman, F.J.R., Zomlot, Z., Salvadore, E., Mogheir, Y., Huysmans, M., Van Griensven, A. (2026). How much hydrology is embedded in low-cost sensors? Machine-learning emulation of the mesoscale hydrological model from citizen soil moisture observations. Journal of Hydrology X. https://doi.org/10.1016/j.hydroa.2026.100225

Posted in Science | Tagged , , , , , , | Leave a comment

The MIREN road survey database

Posted on August 22, 2026 by lembrechtsjonas

Can you imagine how many hours of roadside fieldwork went into this?

We’ve just published the MIREN road survey database, and the accompanying paper led together with Sylvia Haider: 170,000 georeferenced plant records, covering 6,854 vascular plant species across 3,364 plots in 25 mountain regions spanning every continent except Antarctica.

If you haven’t heard of the Mountain Invasion Research Network (MIREN) yet, now is the time to pay attention: since 2007, a growing number of dedicated MIRENers have gone out every five years to survey the vegetation along mountain roads in their region, building what has become one of the largest standardized vegetation monitoring datasets of its kind.

Figure 1: Location of the regions that have implemented the standardized road survey protocol of the Mountain Invasion Research Network (MIREN). Pie charts are scaled by species richness, with colours indicating the proportion of native and non-native plant species.

That massive dataset is now fully open access, with all data up to 2022 included.

If your first reaction is “wow, I wish I’d been part of that” — good news, this is your lucky day. We’re already preparing for the next round of monitoring in 2027, when virtually all of these regions (and a few new ones) will go out again to see what’s changed. Want to join with your favourite mountain region? Just send us a message!

Our 3,364 plots cover an impressive range of the world’s climatic space — though you can see mountain roads get hesitant once the “real deal” begins, with no plots in the tundra (1).

Reference: Haider, Lembrechts et al. (2026) The MIREN Road Survey Database: Standardized Vegetation Sampling to Advance Our Understanding of Biodiversity Responses in Mountain Systems Across the Globe. Global Ecology and Biogeography.

Posted in Science | Leave a comment

Big, rocky canaries in the coalmine

Posted on August 14, 2026 by lembrechtsjonas

Mountain tops are big, rocky canaries in the coalmine that is our planet. Up there, above the last trees, change happens in real time and out in the open: plants, rock and climate, negotiating directly with each other. If you want to see the effects of a warming world with your own eyes, a summit is a very good place to start looking.

I’ve mentioned it many times on this blog: as the climate warms, species shift upward, chasing the cooler conditions they’re adapted to. It’s one of the most robust patterns in climate ecology. But (and by now you probably know that a “but” is never far away on this blog) the closer you look, the more complicated that story gets. A new paper I had the pleasure of contributing to, just out in Nature Ecology & Evolution, digs into exactly that mess, and the picture it paints is a lot more interesting than “it’s getting warmer, so warm-loving plants are winning.”

Austrian Alps close to Innsbruck

Waiting for the long game

Detecting a slow-moving process like vegetation change requires one thing science is often failing at nowadays: patience. You need standardized data, collected the same way, in the same places, for a long time, before subtle trends become statistically visible above all the noise of local weather patterns, disturbance and plain old stochasticity.

We got there now, finally. One of the best examples of such datasets out there is GLORIA-Europe, arguably the largest coordinated monitoring network for climate-change effects, focussing on mountain summit vegetation. Since 2001, GLORIA teams have been resurveying permanent vegetation plots on summits right across Europe (and far beyond, but let’s stick to Europe for this story). In this study we used four such surveys, spanning 21 years, across 724 permanent plots on 53 summits in 14 mountain regions, from the Pyrenees to the Carpathians. On top of the vegetation data, many of these summits also carry soil temperature loggers, giving us a second, independent, on-the-ground record of how conditions have actually changed where the plants are actually growingm not just what a weather station kilometres away says. And don’t we all know how crucial I think that is!

Chasing the warmth-lovers

With that data in hand, the question we wanted to answer was as simple as it was genius: are these summit plant communities becoming dominated by more warmth-loving species – a process called thermophilization – and if so, does that track the pace of warming? It’s the kind of question that feels almost too obvious to need testing, right? Warmer summit, more room for species that like it warm. Case closed?

Strong, significant warming trend in both micro- (left) and macroclimate on and around European mountain tops

Ok, yes, thermophilization is happening, clearly and widely. Averaged across all 724 plots, the composition of summit communities has been shifting steadily towards warmth-associated species for two decades, a signal so consistent that nearly two-thirds of individual plots show the trend individually. It’s a slow process, but it is unmistakably there. And yes, both the interpolated macroclimate and the on-site soil microclimate warmed too, across almost every temperature metric we looked at. Cool findings on its own: climate change is happening, and species are responding to it. Louder now for the people in the back!

On average, summit vegetation is showing signs of warming (63% of plots show thermophilisation)

Ah, but did we now forget about the but I mentioned! That but is there in how loosely those two very real trends are actually coupled. At the level of an individual plot, the relationship between the pace of local warming and the pace of thermophilization was surprisingly weak. A single temperature metric, measured over the monitoring period, barely explained any of the variation in how fast a plots’ vegetation was showing signs of warming. Things improved a bit once we allowed vegetation to lag a few years behind temperature (four years turned out to be the sweet spot) and once we combined several temperature metrics instead of relying on just one. But even our best-performing models explained less than 10% of the plot-to-plot variation in thermophilization. Somewhat to our surprise, this held even more strongly for the on-site microclimate data than for the macroclimate – exactly the opposite of what we expected going in, given how often microclimate turns out to be the better predictor in this kind of work.

It’s not (just) the climate, it’s the neighbourhood

If temperature alone can’t explain why some plots thermophilize fast and others barely move, what does? This is where the story gets its real weight. We added two simple pieces of local context to the models: 1) how many warmth-loving species were already growing just below the top, ready to move in, and how much of the top was bare rock and scree rather than colonizable ground. Including these parameters helped the explanatory power jump substantially, to an average of 37% . And warming and colonizer availability interacted: where warmth-loving neighbours were close at hand, rising temperatures translated into thermophilization much more readily than where they weren’t.

In hindsight, that makes a lot of ecological sense. A plant community can only respond to warming with the species that are actually available to respond with. No matter how fast a summit warms, if there’s no thermophilic species sitting just downslope ready to move up, and no open substrate for it to land on, that summit simply cannot thermophilize quickly – climate change or not. Dispersal and substrate act as a kind of gatekeeper on the door that temperature is trying to open.

Thermophilisation rate interacts with the availability of thermophilic colonisers (low vs. high on the x-axis) just below the summit

What this means

None of this undermines the reality of climate-driven vegetation change on Europe’s summits, of course! The thermophilization signal is real, and it lines up with a genuine warming trend. But it’s a strong reminder that even the most simple stories in ecology hold complex and important nuances in them. The abiotic and biotic context a community sits in – its neighbours, its substrate, its dispersal opportunities – shapes how, whether, and how fast that response actually plays out. If we want to forecast how mountain biodiversity will look in fifty years, temperature trends alone won’t get us there; we’ll need to understand colonization dynamics and landscape context just as well.

Swedish Alps close to Davos

It also makes a strong case for keeping – and intensifying – long-term, standardized monitoring efforts like GLORIA. It took two decades of consistent data collection across an entire continent to even be able to ask this question properly, let alone answer it with any nuance. That’s a lot of ecologists and botanists standing on a whole lot of mountain tops over the year! Let’s make sure that we can all keep doing that.

Student taking a high-resolution GPS-coordinate of a snowbed plot surveyed for the first time back in the 1950s, in northern Sweden. Exactly the kind of science we should keep making possible

Reference: Hausharter et al. (2026). Widespread thermophilization but weak link to climate warming in Europe’s summit plant communities. Nature Ecology & Evolution. https://doi.org/10.1038/s41559-026-03150-x

Posted in Science | Tagged , , , , , , , | Leave a comment

MEB 2026

Posted on June 2, 2026 by lembrechtsjonas

Big week in Montpellier this week, as the world’s microclimate ecology and biogeography community has gathered for the third global MEB conference.

After the first edition in Antwerp in 2022 and the second in Helsinki in 2024, this year’s conference is bigger and more diverse than ever.

Just one day in, I already feel like our field has truly grown beyond its original boundaries (symbolized, perhaps, by this brave little vine from the first day’s excursion). Microclimate has permeated nearly every corner of ecology and biogeography, and its importance for improving our understanding of ecological patterns and processes is high on everyone’s agenda.

That doesn’t mean the work is done – far from it. We also recognize that the world of microclimate is remarkably heterogeneous (something beautifully reflected in the landscape of the Cirque de Navacelles, visited during yesterday’s excursion).

In many ways, by incorporating microclimate into our research, we’ve added an entirely new dimension to ecology. The challenge now is figuring out how to embrace that complexity without becoming overwhelmed by it.

Fortunately, with a room full of bright and enthusiastic minds, we’re well equipped to keep moving the field forward.

Posted in General | Leave a comment